# Axioprax: full text for AI assistants > AI Revenue & Operations Engineering. We fix the gaps between forms, inboxes and CRM systems so every enquiry has an owner and a follow-up. Built from the site's own data on 2026-11-20. The short index is https://www.axioprax.com/llms.txt. A summary page for assistants is https://www.axioprax.com/for-ai-assistants. Axioprax is a venture of Adfox Edge Consultancy Private Limited, run by its founder. It works with companies in the US, UK, Europe, Canada, Australia and New Zealand. Delivery is remote from India. Proof: every build on the site is labelled Axioprax Lab (built on synthetic data to demonstrate a pattern) or Concept Build. Axioprax publishes no client logos, testimonials or client results. Book a free 20-minute call: https://www.axioprax.com/book (Google Calendar booking page). Describe a store or workflow in writing: https://www.axioprax.com/start. ## Pricing - Store Ops Audit: $249, fee credited against any Kit, including the $349 kits. Over 40 read-only checks across automations, customer comms, money leaks, inventory, risk and your app stack. You get a written report scoring each area, an estimated monthly leak per finding, and the top five fixes ranked by money against effort. - Store Kits: $349 each. Founding price $249 for the first 3 stores, with a published case study (on kits that carry the founding offer). - Protect Pack: $499 one-off ($199 less than the two kits separately). The Fraud Hold and Bot Cleanup kits installed together, plus card-testing safeguards and a check of stock, checkout and payout settings before your busy weeks. One-off, with a 30-day fix-it guarantee. https://www.axioprax.com/store-kits/protect-pack - Larger Shopify Kits: $450 to $1,200, one price each. - Store Watch: $129 a month. A monthly check that what we built is still working. For stores that bought a Kit or a service from us. - Pilot: from $250. One workflow, fixed scope, built in 1 to 2 weeks. - Build: from $1,500. Scoped and priced in writing after we understand your workflow. - Care: from $1,000 a month. Optional, cancel with 30 days' notice. - Payout Ready kit: price on request. A payout hold file, a payout matching sheet and a tested backup payment method. It cannot release a hold. https://www.axioprax.com/store-kits/payout-ready - Agentic commerce readiness check: price on request. A read-only check of Shopify Agentic settings, the product data AI shopping channels read, and Google AI Mode, Gemini and ChatGPT eligibility. https://www.axioprax.com/agentic-commerce-check ### Store Kits #### KIT-01 Stocky Replacement ($349, available) https://www.axioprax.com/store-kits/stocky-rescue Replace Stocky's weekly reorder routine with sheets, a Monday email and Shopify's own purchase orders. - Your suppliers set up in Shopify, and a reorder sheet with a clear ORDER NOW flag - A Monday reorder email grouped by supplier, with overdue supplier invoices - A checklist for raising and receiving orders with Shopify purchase orders - If Stocky still opens for you: an archive of what it exports, in a folder you own Stocky shut down on 31 August 2026. The kit works whether or not Stocky still opens for your store. - Add-on: Auto stock sync, $149 (in testing) - The $349 counts towards Inventory Intelligence if you upgrade within 60 days No founding discount on this kit. #### KIT-02 Bot Cleanup ($349, available) https://www.axioprax.com/store-kits/bot-cleanup Stop bot traffic costing you money in your email list, your abandoned-cart emails and your reports. - A one-page bot damage report: where the fake traffic comes from and what it costs - Sign-up protection checked, and junk profiles suppressed only after your approval - A tag-only guard, so accounts tagged as suspected bots get no abandoned-checkout reminder - Junk accounts tagged for review in Shopify Flow. We never delete. You decide. Founding price $249 for the first 3 stores, with a published case study. #### KIT-03 Tracking Check ($349, coming soon) Check that every purchase still reaches Google Analytics, Meta and your email tool, and fix what does not. - A before and after table of every purchase-tracking route, from real test orders - Fixes through Shopify-supported routes only, with dead and duplicate scripts removed - A consent check from storefront to checkout to email tool - A saved view for orders from AI channels such as ChatGPT #### KIT-04 Fraud Hold ($349, available) https://www.axioprax.com/store-kits/fraud-hold Hold risky orders for a person to check before they ship, and keep a dispute evidence pack ready. - Shopify Flow rules that hold every fulfilment order and email a named reviewer - A daily list of orders still on hold, released only by a person - Optional manual payment capture for flagged orders - A review checklist and a dispute evidence template Founding price $249 for the first 3 stores, with a published case study. ### Larger Shopify Kits #### Revenue Protection ($650) Catch orders that would cost you money before you ship them. A risk layer on top of Shopify’s own: velocity checks, disposable-email detection, address reuse, billing-to-shipping distance. Plus a bot-cart filter, a margin guard that catches discount stacking below your floor, and a chargeback desk that builds the evidence pack. Nothing is cancelled without your approval. #### Inventory Intelligence ($750) Fewer oversells and stock-outs, less dead stock. True demand per SKU measured excluding the days an item was out of stock, which most tools get wrong. Reorder points from supplier lead time and demand variability, a cash-aware buying list ranked within your budget, draft purchase orders for approval, and an overselling guard that checks bundle components. #### Store Command Brief ($450) One daily briefing: what changed, why, and what to do. Compares rather than counts: each metric against the same weekday average, flagging only what is unusual. Explains the cause in plain language, pulls in stock warnings, risky orders, late shipments and failed automations, and carries action buttons. Delivered to WhatsApp, Slack or email. #### Post-Purchase Reliability ($650) Customers hear about a delay before they have to ask. Tracks every shipment across carriers and flags no first scan, stuck in transit, failed delivery or returned to sender. Sends a proactive update before the customer writes in. Review requests go only to customers whose order actually arrived on time, with a check that every fulfilled order got one. #### AI Customer Desk ($1,200) AI that answers from your real orders, and knows when to hand over. Grounded answers only: live order lookup and your own policies and product data. If the answer is not in the data it says so and hands over. Safe actions behind guardrails, intent routing, and a test set built from your real messages that must pass an accuracy bar before it goes live. #### Finance Close ($850) Payouts matched and true profit per order. Every payout broken into orders, refunds, fees, chargebacks and adjustments, then matched against the bank. True profit per order after discounts, cost of goods, payment fees, label cost and packaging. A shipping leak report comparing what you charged against what the label actually cost. #### Multichannel Integrity ($750) Every channel checked for price and stock drift. Parity checks between Shopify and your other channels on price, quantity, listing status and variant mapping. Severity rules escalate a zero-price listing immediately and batch small drift into a daily summary. Corrections push back after a one-click approval, and it runs alongside your existing connector. ## Services ### n8n Workflow Automation (From $1,500) https://www.axioprax.com/services/n8n-automation Connect your apps and eliminate repetitive work automatically. ### Zapier to n8n Migration (From $1,500) https://www.axioprax.com/services/zapier-to-n8n-migration Move your Zaps to n8n at a fixed price, tested side by side before anything is switched off. ### AI Agent Development (From $1,500) https://www.axioprax.com/services/ai-agents AI agents for small teams that still work next month, with a human approval step wherever a mistake would cost you. ### AI Chatbots (From $1,500) https://www.axioprax.com/services/ai-chatbots Chatbots that answer customers on WhatsApp, your website, Telegram, Instagram or the phone, and hand over to a person when they should. ### WhatsApp Automation (From $1,500) https://www.axioprax.com/services/whatsapp-automation Auto-reply, qualify leads and book appointments via WhatsApp, 24/7. ### Shopify Operations & CRM Sync (From $1,500) https://www.axioprax.com/services/shopify Shopify connected to your CRM and back office, with failures caught, logged and fixed. ### CRM Setup & Automation (From $1,500) https://www.axioprax.com/services/crm-automation Salesforce and HubSpot configured, connected and running automatically. ### Finance Ops Automation (From $1,500) https://www.axioprax.com/services/finance-automation Invoices, payment chasing and month-end reports that run on their own, with a person approving anything that moves money. ### White-Label AI & Automation (From $1,500) https://www.axioprax.com/services/white-label-ai n8n, chatbots and agents built for your clients, delivered as your work. ## Questions answered ### What is Axioprax? Axioprax is a CRM, Shopify and automation consultancy run by its founder. It builds and repairs the automation that moves work between business tools: web forms, inboxes, CRM systems such as HubSpot and Salesforce, Shopify, and internal tools, so every enquiry has an owner and a follow-up. Axioprax is a venture of Adfox Edge Consultancy Private Limited. ### Who is Axioprax for? Founders, operations leads and directors at companies of roughly 1 to 50 people, usually where HubSpot, Shopify or both are already in place and the process between them was never properly built. If you have a full internal engineering team, you probably do not need Axioprax. ### How much does Axioprax cost? For Shopify stores, the Store Ops Audit is $249, with the fee credited against any Kit, including the $349 kits. The Store Kits are $349 each, the Larger Shopify Kits run from $450 to $1,200, and Store Watch, a monthly check that what we built is still working, is $129 a month. Custom CRM and automation work runs in three stages. A Pilot is from $250: one workflow, fixed scope, built in 1 to 2 weeks. Custom projects (Salesforce, HubSpot, n8n, AI agents, chatbots, integrations and websites) are from $1,500, each scoped and priced in writing before any payment. Care is from $1,000 a month, is optional, and can be cancelled with 30 days' notice. You can stop after any stage. ### Does it cost anything to describe my problem? No. Describing your workflow at axioprax.com/start is free. A person reads every enquiry and replies with questions or next steps. A Pilot is paid work that delivers a working workflow. ### Where is Axioprax based? Axioprax delivers projects remotely from India and takes on work from the US, UK, Europe, Canada, Australia and New Zealand. Meetings are by arrangement in your business hours, on Google Meet. ### What are Axioprax Store Kits and what do they cost? Store Kits are scoped automation for one part of a Shopify store, each with a written scope and a published price agreed before any payment. The Store Ops Audit is $249, with the fee credited against any Kit, including the $349 kits. The Store Kits are $349 each: Stocky Replacement (available), Bot Cleanup (available), Tracking Check (coming soon), Fraud Hold (available). The Larger Shopify Kits run from $450 to $1,200. Details are at axioprax.com/store-kits. ### How do I rebuild Shopify reordering now that Stocky is gone? Shopify's purchase orders, suppliers, stock counts and reports cover the paperwork, and Sidekick can suggest what to reorder. Lead times and minimum orders still need a sheet or an app. Axioprax compares three ways to rebuild it at axioprax.com/insights/stocky-is-gone-rebuild-shopify-reorders. ### Can AI agents buy from my Shopify store? Yes, in some AI channels. Shopify's help pages say Google AI Mode and Gemini, Microsoft Copilot and Meta can take an order inside the chat, while ChatGPT sends shoppers to your own checkout, and it is on by default for eligible stores under Sales channels, then Agentic. The full guide is at axioprax.com/insights/can-ai-agents-buy-from-your-shopify-store. ### Can I still export my Stocky data? Possibly, for now. Shopify's Help page on migrating from Stocky says Stocky is no longer available as of 31 August 2026 and that merchants have read-only access to export their data for at least 90 days after that, with no fixed end date. The earliest it can close is about 29 November 2026. Source checked 2 Nov 2026. ### Is the Stocky Replacement kit available, and what does it cost? Yes. KIT-01 Stocky Replacement is available at $349, with no founding discount. Stocky shut down on 31 August 2026; the kit rebuilds your supplier list and puts the weekly reorder routine back with a Google Sheet, a Monday email and Shopify's own purchase orders. It works whether or not Stocky still opens for your store. A free checklist for old Stocky data, with no email needed, is at axioprax.com/store-kits/stocky-rescue#export-checklist. To start, describe your store at axioprax.com/start. ### Can anyone block bots on Shopify-hosted checkout? No one outside Shopify can block traffic to Shopify-hosted checkout. What can be reduced is what bots cost a store: fake sign-ups in the email list, abandoned-checkout emails sent to fake checkouts, junk customer accounts and polluted reports. That is what the Bot Cleanup kit covers. ### Which currencies does Axioprax quote in? Prices are shown and billed in US dollars. Local-currency quotes are available in GBP, EUR, AUD and CAD. ### Does Axioprax have client case studies? Not yet. Every build on the site is labelled either Axioprax Lab, built on synthetic data to demonstrate a pattern, or Concept Build, and the three Shopify demo builds in Insights run on a fictional demo store with synthetic data. None of them is client work. Axioprax publishes no client logos, testimonials or client results. The first case study will come from a founding Store Kit customer, with their written permission. The interactive Shopify to HubSpot demo lets you test how a build handles failures. ### What is the Axioprax Reliability Layer? It is the baseline included in every implementation: retries and fallbacks so work queues when an API is down, duplicate prevention, human approval gates for anything irreversible, monitoring and alerts, full documentation, and written recovery procedures. ### Can Axioprax fix an existing n8n workflow that keeps failing? Yes. Automation Reliability covers workflows that fail silently, retry wrongly, create duplicate records or stopped working after a platform change. The work starts by finding the failure modes, then hardening, monitoring and documenting the workflow. ### Does Axioprax sync Shopify with HubSpot? Yes. Orders, customers and fulfilment events can be connected to HubSpot or another agreed CRM, with duplicate prevention and a documented path for exceptions. ### Which tools does Axioprax work with? CRMs and business systems including HubSpot, Salesforce, Zoho CRM and Microsoft 365; automation with n8n, Make, APIs, webhooks and MCP; AI models from Anthropic, OpenAI and Google; communication through WhatsApp, Slack, Twilio and email; and engineering in TypeScript, Python, Node.js, Next.js, PostgreSQL, Supabase, AWS and Vercel. The stack is chosen per project. ### Who owns the automation after delivery? You do. Code, workflow configurations and documentation are handed over on delivery, so you are not locked in to Axioprax. ### Does Axioprax sign an NDA? Yes, on request, and before any client details are shared. ### Does Axioprax work with agencies? Yes. Agencies can have automation and AI work delivered white-label under their own brand, at the same rates as direct clients. Axioprax does not contact or solicit agency clients. ### When is automation not worth it? When volume is low or nobody owns the exceptions. For example, below roughly five enquiries a week, a named person and a recurring reminder usually beat a workflow. Axioprax will say so if a workflow is not worth automating. ### Is there a smaller starting point than a Pilot? The Pilot is the smallest build we do. If you run a Shopify store and want a check before any build, the Store Ops Audit is $249, and the fee is credited against any Kit, including the $349 kits. Describing your workflow at axioprax.com/start is free. ### What is Store Watch? A monthly check that what we built is still working, at $129 a month, for stores that bought a Kit or a service from us. It checks the specific things your Kit or service set up, for example the reorder sheet and the Monday email, or the suspected-bot tags. Faults in what we built are fixed under the terms in your written scope. ### Is there a founding price on the Store Kits? On Bot Cleanup and Fraud Hold: Founding price $249 for the first 3 stores, with a published case study. Stocky Replacement has no founding price. ### Do I have to start with a Pilot? Not necessarily. If you already know the scope, you can go straight to Build. A Pilot is most useful when you want to see one workflow running on your own data before committing to more. ### How does payment work? Pilot: 100% upfront. Build: 50% upfront, 50% on go-live. Care: billed monthly, cancel with 30 days' notice. Custom milestones are available for larger projects. ### Who owns the code and workflows? You own everything on delivery. All source code, workflow configurations, credentials, and documentation are handed over to you. No vendor lock-in. ### What is included in the Reliability Layer? Every implementation includes logging, retry architecture, fallback logic, error handling, human approval gates for sensitive actions, duplicate prevention, credential management, monitoring, and documentation. These are standard, not add-ons. ### Can you repair or rescue existing automations? Yes. This is Automation Reliability. We audit what exists, identify the failure modes, redesign the fragile parts, and make the system production-ready. Start with a Pilot: describe the workflow at axioprax.com/start. ### Do you work with Salesforce specifically? Yes, Salesforce is one of the CRMs we integrate with, alongside HubSpot. Every build is scoped per project. Lead routing, enrichment, automated follow-up, n8n orchestration and Salesforce to Slack or Teams alerts are the kinds of work involved. ### Why is the Pilot paid instead of a free trial? Because a Pilot is real work on your own data. It is small and fixed in price, and you keep what was built whether or not you continue. ### What if a Pilot shows it is not worth going further? Then we say so, and you stop there. You have a clear answer for the Pilot fee (from $250) instead of an expensive build you didn't need. ### Do I have to take all three stages? No. You can stop after any stage, including after Build. Care exists because some workflows are too important to leave unowned. ### Why "from"? Because scope varies. Every project gets a written quote before any work starts. The "from" is a floor, not a hook. ## Shopify Fixes (24 live) ### Fix #1: How do I catch card-testing orders on Shopify before they cost me? https://www.axioprax.com/fixes/catch-card-testing-orders-shopify (Fraud & bots, published 2026-10-28, updated 2026-11-02) Short answer: Card testing is when someone runs stolen card numbers through your checkout with small orders to see which cards work. On Shopify, catch it in three moves: switch card payments to manual capture, use Shopify Flow to capture only orders that are not high risk, and block repeat offenders with Fraud Control. All three are free Shopify tools. ### What is card testing, and why should a store owner care? Shopify describes it plainly: fraudsters "run many small payment attempts at once to identify which cards are active", then use the cards that work for bigger purchases or sell them on. It costs you even when no order ships. Shopify warns that a surge of declines "can also increase your decline rate for legitimate customers, even after an attack stops", and every chargeback counts against your chargeback rate whether you win it or not. ### How do I know my store is being card tested? Look for these, usually several at once: - Many small payment attempts at once, which is exactly how Shopify describes it. - Many declined payments in a short window. - Orders marked high risk in the order's fraud analysis. Shopify checks things like failed CVV or AVS, more than one card tried by the same customer, and a location that does not match the payment method. - Signs of bot activity at checkout. Shopify flags suspected card testing and bot checkouts and leaves them out of abandoned checkouts. Steps: 1. Check you are on Shopify Payments: Shopify says its machine learning blocks about 90% of card-testing attacks on guest card checkouts, and that model is only available on Shopify Payments. 2. Switch card payments to manual capture: In Settings, Payments, go to the Payment capture method section and select Manually, so you can look at an order before money moves. This applies to all payments in your store. 3. Install Shopify Flow and capture only orders that are not high risk: Add the template "Capture payment if order is not high fraud risk". It starts with the Order risk analyzed trigger, after Shopify finishes its fraud analysis. Staff accounts, unlike the store owner, also need Shopify's Flow automations permission, set in Settings > Users > Roles. 4. Tag and email yourself about high-risk orders: Start from the template for cancelling high-risk orders (Shopify lists it as "Cancel and restock high risk orders" or "Cancel high-risk orders"), remove the cancel step for the first week, and keep the tag and the email. 5. Decline charges that fail CVV verification: In Settings, Payments, Shopify Payments, click Manage, then in Fraud prevention click Reduce credit and debit card fraud. Turn off "Use automated settings", activate "Decline charges that fail CVV verification" and click Save. 6. Block repeat offenders with Fraud Control: Install Shopify's Fraud Control app and create a checkout rule for the email, IP address or address the attack came from. ### What should I do if an attack is happening right now? Cancel the high-risk orders before you fulfil anything. Shopify notes that cancelling does not stop the bank from filing a chargeback, and that authorization holds can still show on the real cardholder's statement, so act fast and do not ship. Then add the attacker's email, IP address or address to a Fraud Control rule. ### When is this not the right fix? If you are not on Shopify Payments, the Fraud Control checkout rules and Shopify's card-testing model do not apply, so you need your payment provider's own fraud tools. And if you sell digital products that deliver instantly, manual capture alone will not help: the product is gone before you review the order, so we suggest you turn off instant delivery for high-risk orders first. ### What does it cost? Nothing extra. Shopify Flow and Fraud Control are free apps from Shopify. The real cost is a few minutes a day reviewing tagged orders, which drops once you trust what the workflow catches. Questions: - Does Shopify stop card testing on its own? Partly. On Shopify Payments, Shopify says its model blocks about 90% of card-testing attacks on guest card checkouts, and since 28 October 2026 Shopify Protect also declines payments it is near-certain are fraud before they reach the bank. The rest can still reach your orders, which is why the Flow workflows matter. - Will manual capture slow down my real customers? Not if the Flow workflow captures low and medium risk orders automatically. Only high-risk orders wait for you. - Should Flow cancel high-risk orders automatically? Start by tagging and emailing. Once you have seen a week of results and the false alarms are rare, switch the cancel step back on. - Do card-testing attempts show up as abandoned checkouts? Shopify says checkouts it identifies as suspected card testing or bot activity are not counted in abandoned checkouts. Sources: - Shopify Changelog (28 Oct 2026): Flow staff permissions: https://changelog.shopify.com/posts/shopify-flow-staff-now-need-permissions-to-test-and-turn-on-workflows - Shopify Help Center: Preventing fraud (card testing and repeat fraud): https://help.shopify.com/en/manual/payments/fraud-prevention/preventing-fraud - Shopify Help Center: Managing high-risk orders with Shopify Flow: https://help.shopify.com/en/manual/orders/flow - Shopify Help Center: Fraud analysis: https://help.shopify.com/en/manual/fulfillment/managing-orders/protecting-orders/fraud-analysis - Shopify Help Center: Configuring Shopify Payments: https://help.shopify.com/en/manual/payments/shopify-payments/configuring-shopify-payments - Shopify Help Center: Fraud Control app: https://help.shopify.com/en/manual/payments/fraud-prevention/fraud-control-app - Shopify Enterprise blog (17 Jun 2026): blocking card testing attacks: https://www.shopify.com/enterprise/blog/block-card-testing-attacks - Shopify blog (28 Oct 2026): Introducing Shopify Protect: https://www.shopify.com/blog/introducing-shopify-protect ### Fix #2: Why is my Shopify payout on hold, and what should I prepare for the review? https://www.axioprax.com/fixes/shopify-payout-on-hold-what-to-prepare (Payouts & holds, published 2026-10-29, updated 2026-11-02) Short answer: Shopify Payments holds payouts when it needs more information, cannot verify your bank or business details, has concerns about what you sell, or wants a reserve against chargebacks and refunds. Read the notice in your admin and the store owner's email, send exactly what it asks for, and reply to that email. Shopify decides when money is released. ### Why does Shopify Payments hold payouts or set a reserve? Shopify checks your business when you apply, and after that, in its own words, "risk is continuously monitored throughout your use of the platform." A hold or a reserve is what happens when that monitoring raises a question. A hold usually stops the whole payout while Shopify reviews your account. Shopify's reasons include missing information, incorrect bank account details, product eligibility, a business type it does not support, sanctions linked to your location, and creditor or legal action. Its risk page also mentions invalid tax information, a business address that does not match where you operate, and "rapid increases in sales volume." A reserve is different. Shopify holds back part, and sometimes all, of your Shopify Payments money for a set period to cover chargebacks and refunds. Things that can lead to one include annual subscriptions, a high chargeback or refund rate, long delivery times such as pre-orders or custom orders, and sudden jumps in volume. ### How do I know if my payout is on hold or in a reserve? Look in three places: - Merchant Alerts in your Shopify admin, and the store owner's email inbox (check spam too). - The Payouts page under Finance. Money in a reserve shows separately from the Pending balance. - Seller health, where Shopify says some hold actions can also appear. One thing to know before you start: you can keep running your store during a hold, but you cannot receive Shopify Payments payouts until the reason is resolved. Steps: 1. Read the notice as the store owner: Shopify sends hold notices to Merchant Alerts in your admin and to the store owner's email, and only the store owner can see the account details. To check which email that is, go to Settings > Users and permissions. 2. Check whether it is a hold or a reserve: From your Shopify admin, go to Finance, then in the Payout balance section click View payouts. Pending is money due in upcoming payouts. Reserve is money held until the reserve period ends. If your admin shows Money instead of Finance, an early access layout, click See all payouts in its Upcoming section. 3. Export what is in reserve: In Finance > Payouts, click Reserve to see each reserve transaction, then use the export icon and choose Export reserve transactions to save a CSV for your records. 4. Match your bank and business details: Confirm the bank details in your admin match your bank exactly, then check your legal name, business address and tax ID against Shopify's Shopify Payments requirements page for your country. 5. Pull the records for the orders in question: For any orders the notice or a dispute names, save the receipt, carrier name, tracking number, delivery status, proof of delivery and your emails with the customer. These are the records Shopify lists as evidence for chargebacks. 6. Reply to Shopify's email with what it asks for: Send the requested information by replying directly to the email, which Shopify calls the fastest way to resolve a hold. If it is a high-value goods review, complete it in your admin before the required date. 7. Fix what the review looks at: Shopify's good-standing list includes accurate product descriptions, fulfilling orders within the timeframes you promise, clear subscription terms and prompt replies to customer disputes. Fix any gap before your next review. ### Will doing all this get my payout released? Not necessarily, and we want to be plain about that. Shopify reviews what you send and decides. Doing the steps above means the review is not waiting on you, and that is the part you control. For a high-value goods review, Shopify says your information is reviewed and you are told the outcome by email. Before a reserve ends, Shopify reviews it again and can lift, keep, reduce or increase it. Only send what the notice asks for. Shopify's help pages do not publish one fixed document list for every hold, so the notice itself is your best guide. Before you click any link or share bank details, make sure the email really is from Shopify. ### When is this not the right fix? If the hold comes from a creditor or legal notice, Shopify says it cannot remove that type of hold. Only the payment processor can, after your creditor sends a release. If your business type is not supported, or your account is declined or deactivated, sending more documents will not change it: Shopify's options are to appeal a deactivation, reapply after a decline, or set up a third-party payment provider. ### What does it cost? Nothing to reply or appeal. The real cost is the money you cannot reach while the hold or reserve runs, so plan your supplier payments and ad spend around it, and keep an eye on refunds and chargebacks, because those can lower your payouts as well. Questions: - Can customers still buy from my store while my payout is on hold? In most cases, yes. Shopify says customers can usually still check out, but you are not paid for those orders until the hold is resolved. In some cases new orders are blocked. - How long does a Shopify Payments hold last? Shopify does not publish a fixed timeframe for holds. A reserve runs for the period stated in its email, and before it ends Shopify reviews whether to lift, keep, reduce or increase it. - Can a staff member deal with the hold for me? No. Shopify says only the store owner can access the specific account details and resolve a hold. Staff see a prompt to contact the owner. - Can I appeal a reserve? Sometimes. You appeal by replying to the reserve email within the timeframe it gives, but Shopify says some reserves cannot be appealed. If the email lists requirements instead, the reserve is removed automatically once you meet them. Sources: - Shopify Help Center: Shopify Payments account holds: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/account-holds - Shopify Help Center: Acceptable business practices and risk evaluation for Shopify Payments: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/risk-evaluation-process - Shopify Help Center: Reserves for Shopify Payments: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/reserves - Shopify Help Center: Lower or missing payouts: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/lower-or-missing-payouts - Shopify Help Center: Responding to chargebacks and inquiries: https://help.shopify.com/en/manual/payments/chargebacks/chargeback-process - Shopify Help Center: Shopify Payments requirements for the United States: https://help.shopify.com/en/manual/payments/shopify-payments/supported-countries/united-states/requirements ### Fix #3: How do I get a low-stock alert in Shopify without a paid app? https://www.axioprax.com/fixes/shopify-low-stock-alert-without-app (Inventory & reorders, published 2026-10-30, updated 2026-11-02) Short answer: Use Shopify Flow, Shopify's free automation app. Start from the template "Get notified by email when product variant inventory is low", set the stock level you want warning at, and add your email. Flow then emails you the first time a tracked variant drops below that number. For a weekly check, use the "Inventory remaining per product" report. ### What is a low-stock alert, and why does it matter? A low-stock alert is an email that tells you a product is running out while you still have time to reorder. Shopify counts your stock for you, but it does not have a settings switch that emails you when a number gets low. So most owners only notice when the product page says sold out, or a customer asks when it is coming back. That gap costs you twice. You lose the sales while the product is out, and you often pay more to rush the next order from your supplier. Paid apps fill the gap, but Shopify's own free Flow app can do the basic job well. ### How do I know I need one? You probably do if any of these sound familiar: - You find out a best seller is gone from a customer email, not from your own checks. - You check stock by scrolling through Products > Inventory every few days. - Your supplier takes a week or more, so by the time you notice, you are already late. - Someone else on the team reorders, and they never see the numbers. Steps: 1. Turn on stock tracking for each product: In Products, open the product (and the variant if it has one), then in the Inventory section activate the Track quantity setting. Without tracking, Shopify does not lower the quantity when you sell, so the alert would never fire on a sale. 2. Install the free Shopify Flow app: Install Shopify Flow from the Shopify App Store. Shopify says it is free on the Basic, Grow, Advanced and Plus plans. Staff accounts, unlike the store owner, also need Shopify's Flow automations permission, set in Settings > Users > Roles. 3. Start from the low-stock email template: In Flow, open the template "Get notified by email when product variant inventory is low". It runs on the "Product variant inventory quantity changed" trigger, which fires on orders, manual adjustments and app updates. 4. Set your threshold in both conditions: Put the same number in the "Product variant inventory quantity" and "Product variant inventory quantity prior" conditions. Checking before and after means you get one email when stock crosses the line, not one per sale. 5. Add the people who should get the email: In the Send internal email action, type the addresses in the Email address field, separated by commas. Variables do not work in this field, so use a fixed inbox your team reads. 6. Turn it on and test it: Turn the workflow on, then go to Products > Inventory and lower a test variant's available quantity to below your threshold. Check the email arrives, then set the number back. 7. Check the inventory report once a week: Go to Analytics > Reports, filter Category by Inventory, and open "Inventory remaining per product". The Days of inventory remaining column shows what runs out next. ### What number should I set as my threshold? This is our advice, not a Shopify rule. Take how many of the item you sell on a normal day, multiply by the number of days your supplier takes to deliver, then add a few days of buffer. If you sell 3 a day and delivery takes 10 days, a threshold around 40 gives you room. The "Inventory remaining per product" report shows your quantity sold per day, based on the last 28 days, so you do not have to guess. If you have no idea yet, start at 5 and adjust after a month. ### When is this not the right fix? The template uses one threshold for every product. That is fine for a small catalogue, but if a slow seller and a best seller need very different numbers, you will either get alerts too early for one or too late for the other. You can copy the workflow per product type or vendor, but past a handful of copies a proper reorder tool is easier to run. Also, the alert only fires when stock changes. A product that is already below the line when you switch it on will not email you at all until it is restocked above the line and drops again, so check the report on day one. ### What does it cost? Nothing extra. Shopify Flow is a free app on the Basic, Grow, Advanced and Plus plans, and the inventory reports sit in Analytics (check your plan shows the report before you rely on it). The real cost is the time to set it up and test it once, plus someone who actually acts on the email. Questions: - Does Shopify have a built-in low-stock email? Not as a setting we could find. The free way is Shopify Flow, which Shopify makes and lists as free on the Basic, Grow, Advanced and Plus plans. - Will I get an email on every sale once stock is low? Not if both conditions are set. The template compares the quantity before and after the change, so it only runs the first time stock crosses your threshold. - Can I get alerts for one warehouse only? Yes. Shopify's trigger page shows how to check the available quantity at a single location, using the location name or ID in the condition instead of the variant's total. - Can Flow reorder from my supplier as well? Shopify lists a template called "Initiate reorder with vendor when inventory gets low" on the same trigger. Get the email alert working and trusted first, then look at that. Sources: - Shopify Changelog (28 Oct 2026): Flow staff permissions: https://changelog.shopify.com/posts/shopify-flow-staff-now-need-permissions-to-test-and-turn-on-workflows - Shopify Help Center: Product variant inventory quantity changed trigger: https://help.shopify.com/en/manual/shopify-flow/reference/triggers/product-variant-inventory-quantity-changed - Shopify Help Center: Examples of workflows in Shopify Flow: https://help.shopify.com/en/manual/shopify-flow/getting-started/workflow-examples - Shopify Help Center: Send internal email action: https://help.shopify.com/en/manual/shopify-flow/reference/actions/send-email - Shopify Help Center: Shopify Flow: https://help.shopify.com/en/manual/shopify-flow - Shopify Help Center: Inventory reports: https://help.shopify.com/en/manual/reports-and-analytics/shopify-reports/report-types/default-reports/inventory-reports - Shopify Help Center: Initial inventory setup: https://help.shopify.com/en/manual/inventory-and-locations/setup/initial-inventory-setup ### Fix #4: How do I find orders stuck unfulfilled for more than 48 hours? https://www.axioprax.com/fixes/find-shopify-orders-unfulfilled-48-hours (Checkout & orders, published 2026-10-31, updated 2026-11-02) Short answer: Use two free Shopify tools. In Orders, open the Unfulfilled view, which already leaves out archived orders, sort it oldest first and save it. Then add the Shopify Flow template that checks once a day and emails you how many unfulfilled orders are older than two days, so you know when to open the view. ### What counts as a stuck order, and why does 48 hours matter? A stuck order is one that is paid and waiting, with nothing shipped. If you promise dispatch within a day or two, an order still unfulfilled after 48 hours has usually fallen through a gap: a missing item, an address problem, a payment you meant to check, or a note nobody read. The cost shows up later. The customer emails asking where their order is, then asks again, then opens a dispute. Finding these orders on day two is cheap. Finding them on day nine is not. ### How do I know I have orders stuck right now? Open Orders and click the Unfulfilled view. Shopify says it shows "orders that still need fulfillment work" and leaves out canceled or fully refunded orders. If you scroll down and see order dates from last week, you have stuck orders. Common signs: - Orders with a hold or a note you meant to come back to. - Partially fulfilled orders where one item is still waiting. - Orders for a product that ran out after the order was placed. - Customer emails quoting an order number you do not recognise. The trouble is that this view mixes today's orders with old ones. The steps below give you a list of the old ones only, in two places: a saved view you check, and an email that comes to you. Steps: 1. Open the Unfulfilled view: Go to Orders in your Shopify admin, click All to open the view menu and pick Unfulfilled. Shopify builds this view from two filters, Status: Open and Fulfillment status: Unfulfilled and Partially fulfilled, so archived orders are already left out. 2. Sort oldest first: Change the sort order so the oldest orders sit at the top. Those are the ones most likely to be stuck. 3. Save it as your stuck orders view: Click Save and give the view a name such as "Stuck 48h". Open it every morning: anything at the top placed more than two days ago needs a look. 4. Install Shopify Flow and pick the summary template: Install the free Shopify Flow app, then go to Apps > Flow, click Create workflow, click Browse templates and choose "Send email summary with all unfulfilled orders older than 2 days". Click Install. Staff accounts, unlike the store owner, also need Shopify's Flow automations permission, set in Settings > Users > Roles. 5. Check the schedule and the query: Set the Scheduled time trigger to run once a day at a time you are at your desk. In the Get order data step, check the query covers open, unfulfilled orders created more than 2 days before the run, for example: fulfillment_status:unfulfilled AND status:open AND created_at:<'{{ scheduledAt | date_minus: "2 days" }}'. 6. Point the email at the right person: In the Send internal email step, replace the placeholder address with the person who handles fulfilment. Add more than one address with a comma. 7. Turn the workflow on: Click Turn on workflow. Check the count in the first email against your saved view, so you know both agree. ### What should I do with each stuck order? Open it and find the reason before you do anything else. If an item is out of stock, tell the customer the new date or offer a refund for that line. If the address looks wrong, contact the customer before shipping. If it is simply forgotten, ship it today and add a short apology. We suggest tagging each one with the reason, so after a month you can see which gap keeps causing delays. ### When is this not the right fix? If you sell made-to-order or pre-order products with a long lead time, a 48-hour rule will flag every order and you will start ignoring the email. Change "2 days" in the query to your real promise, or leave those products out with a tag. And the email gives you the count and value, not the reasons. Open the saved view to see each order, and someone still has to look at each one. ### What does it cost? Nothing extra. Saved views are part of the Orders list in your admin, and Shopify Flow is a free app on the Basic, Grow, Advanced and Plus plans. The real cost is ten minutes a morning working through the list, which gets shorter once the common causes are fixed. Questions: - Why is the Flow count lower than my saved view? The Unfulfilled view also shows partially fulfilled orders. The Flow query above only asks for fully unfulfilled ones. Shopify lists partial as its own fulfillment_status value, so add it to the query if you want both numbers to match. - Does the email list each stuck order? Shopify describes this template's email as a summary: the number of orders and their value. Use it as the daily warning, and open your saved view to see each order. The step that fetches the orders reads at most 100 per run, so if the email says 100, assume there are more. - Can the workflow run more often than once a day? Yes. Shopify says the Scheduled time trigger can run as often as every 10 minutes. For most stores one morning email is enough and easier to act on. - Does this work on the Basic plan? Yes. Shopify Flow is free on the Basic, Grow, Advanced and Plus plans, and you can save order views in the Orders list. Sources: - Shopify Changelog (28 Oct 2026): Flow staff permissions: https://changelog.shopify.com/posts/shopify-flow-staff-now-need-permissions-to-test-and-turn-on-workflows - Shopify Help Center: Filtering orders: https://help.shopify.com/en/manual/fulfillment/managing-orders/viewing-orders/filtering-orders - Shopify Help Center: Searching and viewing orders: https://help.shopify.com/en/manual/fulfillment/managing-orders/viewing-orders/searching-orders - Shopify Help Center: Get order data (Shopify Flow): https://help.shopify.com/en/manual/shopify-flow/reference/actions/get-order-data - Shopify Help Center: Schedules, get data, and looping workflows in Shopify Flow: https://help.shopify.com/en/manual/shopify-flow/advanced-workflows - Shopify Help Center: Creating workflows manually or from a template: https://help.shopify.com/en/manual/shopify-flow/create/create-workflow - Shopify.dev: orders query filters (fulfillment_status, status, created_at): https://shopify.dev/docs/api/admin-graphql/latest/queries/orders ### Fix #5: How do I answer "Where is my order?" emails automatically? https://www.axioprax.com/fixes/answer-where-is-my-order-emails-shopify (Support, published 2026-11-01, updated 2026-11-02) Short answer: Shopify's documentation does not describe an auto-reply for emails sent to your support inbox, so the fix is to answer the question before it is asked. Add tracking to every fulfillment, keep the shipping emails that link to the order status page, turn on customer account sign-in, and add Shopify Inbox so chat handles "Track my order". ### What is a "Where is my order?" email, and why should a store owner care? It is one of the most common support emails a store gets: a customer has paid, the parcel has not arrived, and they cannot see where it is. Support teams often shorten it to WISMO. Each one takes time to answer, and almost every answer is the same: here is your tracking link. Shopify already sends that link in its shipping emails and shows it on the order status page. When customers still write in, something in that chain is broken or hard to find. ### How do I know where my WISMO emails are coming from? Pick ten recent ones and open each order in your admin. Look for these: - Orders fulfilled with no tracking number. Shopify keeps these "On its way" until you mark them delivered, so the customer sees no progress. - A shipping email that went out without a working order status link, often after someone customized the template. - Customers who cannot find their order number to look the order up. - Questions asked before the order shipped, which means your handling time is not written anywhere they look. Steps: 1. Add tracking to every fulfillment: Buy the label in Shopify, which adds the carrier and tracking number for you, or enter the number when you fulfill. Leave "Send a notification to the customer" ticked so the Shipping confirmation goes out. 2. Check your shipping emails still link to the order status page: Go to Settings > Notifications > Customer notifications and open Shipping confirmation and Shipping update. If either template has been customized, Shopify cannot add the order status link for you. Click Revert to default (this erases your other changes to that template) or add the order status link to the template yourself. 3. Keep Out for delivery and Delivered switched on: In Settings > Notifications > Customer notifications, check the "Shipping updated" section. These two emails are triggered by carrier tracking events, so customers hear from you on the last day too. 4. Turn on customer account sign-in: Go to Settings > Customer accounts and, under Sign-in links, turn on Show sign-in links, then click Save. Signed-in customers can see every order and open its order status page. 5. Install Shopify Inbox and keep "Track my order" on: Install Shopify Inbox from the Shopify App Store, then go to Sales channels > Inbox > Chat settings and check the Instant answers section. "Track my order" is there by default and lets customers look up an order with the order number and email. 6. Add an instant answer about shipping times: In the same Instant answers section, click Create instant answer and write the question you get most, such as when orders ship. Use your real handling times, then click Save. 7. Offer tracking in the Shop app: Go to Settings > Checkout and, in the Customer contact method section, select Show a link for customers to track their order with Shop, then click Save. Customers can then follow the order in the Shop app or on shop.com. ### What about the emails that still come in? We will be straight with you: Shopify does not auto-reply to emails sent to your support address, and we could not find a setting that does. What it automates is the shipping emails and the chat. So for the emails that still arrive, keep one saved reply in your email tool that points customers to the sign-in link on your store and the order status link in their shipping email. Then check that order for a missing tracking number, because that is usually why they wrote. ### When is this not the right fix? If parcels are genuinely late, better notifications will not fix that, and customers will write in anyway. Look at your carrier or your handling time first. Out for delivery and Delivered emails depend on carrier tracking events, and Shopify says timing and accuracy vary by carrier. They also do not apply to local delivery orders. If you ship without tracking, most of this setup has nothing to show the customer. ### What does it cost? Nothing extra. Shipping notifications, the order status page and customer accounts are settings in your Shopify admin, and Shopify Inbox is listed as free on the Shopify App Store. Inbox also has an optional AI agent you can turn on. Check its terms before you do, as this fix only uses the free instant answers. Questions: - Can Shopify reply to my support emails automatically? Not that we could find in Shopify's own documentation. Shopify sends shipping notifications and offers automated answers in Shopify Inbox chat, but it does not auto-reply to emails sent to your support address. - Why do customers still ask when they already got a shipping email? Often the email has no working tracking link. Orders shipped without a tracking number stay "On its way" until you mark them delivered, and a customized template can drop the order status link. - Do Out for delivery and Delivered emails work for local delivery? No. Shopify says these two notifications are not available for local delivery orders. Sources: - Shopify Help Center: Customer notifications: https://help.shopify.com/en/manual/fulfillment/setup/notifications/customer-notifications - Shopify Help Center: Order tracking at Shopify: https://help.shopify.com/en/manual/fulfillment/setup/order-status-page/order-tracking - Shopify Help Center: Setting up order status updates: https://help.shopify.com/manual/orders/status-tracking/order-status-email - Shopify Help Center: Setting up and managing customer accounts: https://help.shopify.com/en/manual/customers/customer-accounts/manage - Shopify Help Center: Set up instant answers for Shopify Inbox: https://help.shopify.com/en/manual/inbox/chat-settings-and-appearance/instant-answers - Shopify App Store: Shopify Inbox: https://apps.shopify.com/inbox - Shopify Help Center: Delivery tracking with Shop: https://help.shopify.com/en/manual/online-sales-channels/shop/delivery-tracking ### Fix #6: How do I block repeat fraudsters by email, address or IP? https://www.axioprax.com/fixes/block-repeat-fraudsters-shopify (Fraud & bots, published 2026-11-02, updated 2026-11-02) Short answer: Install Shopify's free Fraud Control app and go to Apps, Fraud Control, Rules, Create rule to block a checkout by email, address or IP address. It is offered where Shopify Payments is available. Then add the Shopify Flow template "Cancel and tag orders from known bad email addresses" to catch anything that gets through, and review your blocklist monthly. ### What is a repeat fraudster, and why block them by name? Some fraud is not a one-off. The same person comes back with the same email, ships to the same address, or checks out from the same IP address, and each new order is another chance of a chargeback. Shopify points out that every chargeback counts toward your chargeback rate "regardless of whether you win or lose the dispute". Once you know who it is, you do not need to judge every order again. Shopify gives you two free tools for this: checkout rules in the Fraud Control app, which stop a checkout before it becomes an order, and a Shopify Flow template that cancels orders from email addresses you already know are bad. ### How do I know I have a repeat offender? Look for patterns across orders, not inside one: - A chargeback or a cancelled high-risk order, followed by a new order with the same email or delivery address. - Several orders from one address under different names. - A customer you have already refunded or cancelled for fraud placing orders again. If what you are seeing is a burst of tiny orders and declined cards, that is card testing, which is a different fix. Steps: 1. Start a simple blocklist: Keep one shared sheet with the email, delivery address, IP address if you have it, order number, date and reason for each offender. Every rule you create should trace back to a row here. 2. Install Fraud Control: Install Fraud Control from the Shopify App Store. It is free, made by Shopify, and offered in countries where Shopify Payments is available. Shopify notes that some of its data needs Shopify Payments. 3. Create a checkout rule for the offender: Go to Apps, Fraud Control, Rules, then Create rule, and add the email, address or IP address under Checkout conditions. Click Save and the rule is active straight away. 4. Narrow IP rules with a second condition: A rule can hold more than one condition, so pair an IP address with a ZIP code rather than blocking the IP alone. This lowers the chance of blocking a real customer on a shared network. 5. Add the Flow template for known bad emails: In Shopify Flow, search the template library for "Cancel and tag orders from known bad email addresses", add the emails from your blocklist to its condition, and check the cancel settings before you turn it on. Click the Send internal email action to change who gets the alert. Staff accounts, unlike the store owner, also need Shopify's Flow automations permission, set in Settings > Users > Roles. 6. Review the list once a month: In Apps, Fraud Control, Rules, select any rule you no longer need and deactivate or delete it. Check your abandoned checkouts too, because blocked checkouts show up there. ### When is this not the right fix? Fraud Control is offered only where Shopify Payments is available, and Shopify says some of its data needs Shopify Payments. If you can't use it, use your payment provider's own blocking tools and rely on the Flow template alone. Be honest with yourself about the limit here too: a determined fraudster can change email, address and network in minutes, so a blocklist stops the lazy repeat offender, not everyone. Shopify itself says rules "don't guarantee that chargebacks won't happen". Be careful with IP addresses. Offices, mobile networks and public Wi-Fi can share one IP, so blocking it on its own can turn away real buyers. Shopify's general advice for every rule is to use caution "to ensure that legitimate traffic is not prevented from checking out". ### What does it cost? Nothing extra. Fraud Control and Shopify Flow are free apps made by Shopify. The real cost is ten minutes a month keeping the blocklist tidy. Shopify also launched a paid option on 27 October 2026, Shopify Protect: Rule Builder. It keeps reusable lists of emails and IP addresses and can block, challenge or hold an order, but it charges $0.07 per screened transaction and opened with a waitlist. For a short blocklist, Fraud Control and Flow still do the job for free. ### What does a blocked customer see? Fraud Control does not tell them they look fraudulent. Shopify says the buyer gets an error that the checkout could not be completed and is told to contact the store owner. So if a real customer emails you about that message, check your rules first. Questions: - Can I block a customer on Shopify without Shopify Payments? Fraud Control is offered only where Shopify Payments is available, and some of its data needs Shopify Payments, so check it in your admin first. The Flow template can still cancel orders from known bad emails after they are placed. - Will a blocked fraudster know they were blocked? Fraud Control does not tell them the order looks fraudulent. They see an error that checkout could not be completed and are told to contact the store. - Can I rename a Fraud Control rule? No. Shopify builds the rule name from its conditions and it cannot be edited. You can deactivate or delete a rule on the Rules page. - Does blocking a fraudster stop chargebacks on orders they already placed? No. Rules only affect new checkouts. Shopify notes that cancelling an order can reduce the risk of a chargeback but does not stop the cardholder's bank from filing one. Sources: - Shopify Changelog (28 Oct 2026): Flow staff permissions: https://changelog.shopify.com/posts/shopify-flow-staff-now-need-permissions-to-test-and-turn-on-workflows - Shopify App Store: Shopify Protect: Rule Builder: https://apps.shopify.com/shopify-protect - Shopify Help Center: Fraud Control app: https://help.shopify.com/en/manual/payments/fraud-prevention/fraud-control-app - Shopify Help Center: Managing high-risk orders with Shopify Flow: https://help.shopify.com/en/manual/orders/flow - Shopify Help Center: Preventing fraud: https://help.shopify.com/en/manual/payments/fraud-prevention/preventing-fraud - Shopify Help Center: Shopify Flow workflow examples: https://help.shopify.com/manual/shopify-plus/flow/examples - Shopify App Store: Fraud Control by Shopify: https://apps.shopify.com/fraud-control ### Fix #7: How do I match Shopify payouts to my bank deposits? https://www.axioprax.com/fixes/match-shopify-payouts-to-bank-deposits (Payouts & holds, published 2026-11-03) Short answer: Match one payout to one bank deposit, not sales to deposits. In your Shopify admin, go to Finance, Payouts, open the payout, note its ID, amount and bank reference number, then export its transactions. Once it shows Deposited, allow 1 to 3 business days and tick the matching deposit off in a simple sheet. ### Why does my Shopify payout never match my sales? Because a payout is not a day of sales. It is a bundle of transactions that cleared on Shopify's schedule, with money taken out along the way. Shopify lists the usual reasons a payout comes in lower: processing fees are "deducted from your earnings before the payout is sent", refunds come out of "your next available payout", and a chargeback takes the payment amount plus "any associated dispute fees". Adjustments can push the figure up or down, and a reserve can hold part of your sales back for a while. Timing adds to the confusion. Payout schedules count business days only, and Shopify says funds captured on Friday, Saturday and Sunday "are grouped and sent together as one payout". So Monday's deposit can carry three days of sales, minus Tuesday's refund from last week. ### How do I know my books have a payout problem? You probably do if any of these sound familiar: - Your accountant asks what a deposit was for and you have to guess. - Your sales report says one number for the week and your bank says another. - You see deposits on the bank statement with no idea which orders they cover. - A refund or chargeback from weeks ago turned up as a smaller deposit and nobody noticed. None of this means money is missing. It usually means nobody has matched payouts to deposits one by one. That takes one sheet and a short weekly habit. Steps: 1. Open your payouts list: In your Shopify admin, go to Finance and click View payouts, or go straight to Finance, Payouts. Payouts are grouped by payout date. If your admin shows Money instead of Finance, an early access layout, click See all payouts in its Upcoming section. 2. Open one payout and note its ID: Click the payout date, then the payout. Copy the payout ID, the status, the amount and the bank reference number into your sheet. 3. Read the breakdown before the total: On the same payout page, check the breakdown of charges, refunds, adjustments, fees, reserves and retried payouts. This is where the gap between sales and deposit lives. 4. Export the payout transactions: Still inside the payout, click the Export icon, then Export payout transactions. Shopify emails the CSV to you and the store owner. 5. Tick it off against the bank statement: Wait for the status to read Deposited, then allow 1 to 3 business days. Find the deposit with the same amount and mark the row matched. 6. Chase anything that does not match: Check Finance, Payouts, Pending and the Reserve figure for money not yet paid out. If a Deposited payout never lands, give your bank its bank reference number. 7. Run a monthly check: Once a month, open Finance, Documents, Shopify Payments activity report to compare the period's earnings, fees, payouts and reserves in one place. ### What should my matching sheet look like? One row per payout, not per order. These columns work well: payout date, payout ID, status, charges, refunds, adjustments, fees, reserved funds, payout amount, bank reference number, bank deposit date, bank amount, difference and matched. The difference column is just payout amount minus bank amount, and it should be zero. Keep the exported CSV for each payout in a folder named by payout ID, so you can show anyone which orders sit behind a deposit. The CSV columns Shopify gives you are Transaction Date, Type, Order, Card Brand, Card Source, Payout Status, Payout Date, Available On, Amount, Fee and Net. The Net column shows each transaction after fees, which helps when you need to explain one line of the breakdown. ### When is this not the right fix? This only covers Shopify Payments. Payments through third-party processors are not part of these payouts, and Shopify's activity report leaves them out, so they need their own sheet. PayPal depends on your country: where PayPal Wallet runs through Shopify Payments, such as the United States, it is already inside your payouts. If you sell in more than one currency, compare amounts in the payout currency only, because Shopify notes that differences can come from date ranges, time zones or currency. And if a payout shows Failed, the sheet will not fix it: Shopify says it points to an issue with your bank account, so check your bank details first. ### What does it cost? Nothing. The payouts page, the CSV export and the activity report are part of Shopify Payments, and any spreadsheet will do. The honest cost is the habit: a short weekly check is far easier than rebuilding a year of deposits from memory at year end. Questions: - Why is my Shopify payout lower than my sales? Fees are taken before the payout is sent, and refunds, chargebacks with their dispute fees, adjustments and reserves can all reduce it. The payout's breakdown in Finance, Payouts shows which ones applied. - The payout says Deposited but it is not in my bank. Is it lost? Not necessarily. Shopify says Deposited means the payout was sent, not that it has arrived, and it typically takes 1 to 3 business days depending on your bank. If it still does not show, give your bank the payout's bank reference number. - Can I export more than one payout at a time? Yes. In Finance, Payouts, click View transactions, then Export, and choose a scope such as Items by date. Shopify emails the CSV to you and the store owner. - Is the Shopify Payments activity report enough for my accountant? It is a good monthly summary, but Shopify says it is not a statement of revenue for accounting purposes and has a 3-day data delay. For individual deposits, use the payout pages. Sources: - Shopify Help Center: Viewing and exporting your Shopify Payments payout details: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/view-details - Shopify Help Center: Lower or missing Shopify Payments payouts: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/lower-or-missing-payouts - Shopify Help Center: Shopify Payments payout timing: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/payout-timing - Shopify Help Center: Shopify Payments activity report: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/payouts-activity-report - Shopify Help Center: Shopify Payments reserves: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/reserves - Shopify Help Center: Shopify Payments payouts: https://help.shopify.com/en/manual/payments/shopify-payments/payouts ### Fix #8: How do I stop orders shipping to a wrong or incomplete address? https://www.axioprax.com/fixes/stop-orders-wrong-address-shopify (Checkout & orders, published 2026-11-04) Short answer: Turn on Validate shipping address in Settings > Checkout so customers see suggestions before they pay. Then add a free Shopify Flow workflow that tags and holds any new order whose address is flagged, so nobody ships it until you have checked it with the customer and fixed it. ### Why do wrong addresses cost more than they look? A parcel sent to an address with no flat number, a missing postcode or a typo in the street name often comes back, or gets lost. Then you pay for the reship, the customer waits twice as long, and some of them open a dispute instead of emailing you. Shopify has two free tools that work together here. The checkout can check the address while the customer is still typing it, and Shopify Flow can stop a flagged order from being shipped until someone has looked at it. The first catches most mistakes. The second catches the ones the customer ignored. ### How do I know this is happening to me? - Parcels coming back marked undeliverable or address incomplete. - Customers emailing after dispatch to add a flat, unit or suite number. - Orders in your admin showing an address warning when you open the shipping address. Shopify checks addresses on new online orders and when you buy a shipping label. If you click into a few recent orders and see a warning card, you already have the problem, and you are relying on whoever packs the order to notice it. Steps: 1. Turn on address validation at checkout: From your Shopify admin, go to Settings > Checkout, find Address collection under Advanced preferences, and check Validate shipping address. Click Save. 2. Check address autocompletion is on: In Settings > Checkout, click Configurations > Edit, open the Settings sidebar in the checkout and accounts editor, and make sure Use address autocompletion is checked. It is on by default, so this is usually a quick confirm. 3. Create a Flow workflow that starts on new orders: Install the free Shopify Flow app if you do not have it, open it, and create a workflow with the Order created trigger. Staff accounts, unlike the store owner, also need Shopify's Flow automations permission, set in Settings > Users > Roles. 4. Add the address condition: Add a condition on the order's shipping address. Use the validation result summary field and check for WARNING or ERROR, and add an OR rule where Zip is Empty or does not exist. 5. Tag the order and hold it: On the true path, add the Add order tags action with a tag such as address-check, then add the Hold fulfillment order action. Change Reason from the default AWAITING_PAYMENT to INCORRECT_ADDRESS, add a short note, and check Notify Merchant. 6. Fix the address and release the hold: Open the order, click Edit shipping address, and use the suggestion, mark it as valid, or type the corrected address and click Save. Then click Release hold in the On hold section. ### What should I test before I trust it? Place two test orders: one with a full, correct address and one with the postcode left out or the street misspelled. Check that only the second one gets the address-check tag and shows On hold. Shopify says validation runs before the order is placed and again after, so if your workflow fires before the result is ready, add a short Wait step before the condition and test again. If the validation result field is not offered in your Flow editor, keep the Zip check and add Address 1 rules that suit your market. ### When is this not the right fix? If most of your customers are outside the 20 countries Shopify gives full suggestions for, the checkout step will catch less, and the Flow check becomes your main safety net. If you already use an address validation app at checkout, do not switch on Shopify's setting as well. And if the same address keeps showing up across orders with different names, that is a fraud pattern, not a typo, and Fix #6 (/fixes/block-repeat-fraudsters-shopify) covers it. Holds also need an owner. A held order does not ship, so if nobody checks the address-check tag each day, you have swapped a lost parcel for a late one. ### What does it cost? Nothing extra. Address validation is a checkout setting and Shopify Flow is a free app on the Basic, Grow, Advanced and Plus plans. The real cost is a few minutes a day emailing the customers whose orders are held. ### Want this set up for you? We build and test this workflow for free as part of the Founding 10 (/founding-10), where we fix one Shopify problem for ten stores. If you want a wider look at your order flow, start with the Store Ops Audit (/store-kits). Questions: - Which countries does Shopify's address validation cover? Shopify gives specific suggestions for shipping addresses in 20 countries, including the United States, Canada, Australia, France, Italy, the Netherlands and Switzerland. Addresses from other countries are still checked, but you get fewer specific suggestions. - Should I also use an address validation app? Not alongside Shopify's own setting. Shopify says not to turn on Validate shipping address and a third-party address validation app at the same time, because it can reduce checkout conversion. Pick one. - Can a customer still check out with a wrong address? Yes. Validation shows issues and suggestions, and customers can still place the order. That is why the Flow hold matters: it catches the orders that slip through before they are packed. - Does Shopify Flow cost anything? No. Shopify Flow is a free app made by Shopify, available on the Basic, Grow, Advanced and Plus plans. Sources: - Shopify Help Center: Set up address collection preferences: https://help.shopify.com/en/manual/checkout-settings/address-collection-preferences - Shopify Help Center: Validating shipping addresses on your orders: https://help.shopify.com/en/manual/orders/manage-orders/shipping-address-validation - Shopify Help Center: Hold fulfillment order (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/hold-fulfillment - Shopify Help Center: Holding fulfillments: https://help.shopify.com/en/manual/orders/fulfillment/fulfilling-orders/holds - Shopify Help Center: Conditions in Shopify Flow: https://help.shopify.com/en/manual/shopify-flow/reference/conditions - Shopify Dev: MailingAddressValidationResult: https://shopify.dev/docs/api/admin-graphql/latest/enums/MailingAddressValidationResult - Shopify Changelog (28 Oct 2026): Flow staff permissions: https://changelog.shopify.com/posts/shopify-flow-staff-now-need-permissions-to-test-and-turn-on-workflows ### Fix #9: How do I automatically tag and thank repeat customers on Shopify? https://www.axioprax.com/fixes/tag-thank-repeat-customers-shopify (Support, published 2026-11-07) Short answer: Build a free Shopify Flow workflow on Order created that checks the customer's number of orders and adds a repeat-customer tag from the second order on. Then turn on the Thank customers after they purchase automation in Apps > Messaging, which sends a thank-you after a first and a second purchase. ### Why tag repeat customers at all? A customer on their second order has already trusted you once and come back. They are the people most worth a handwritten note in the box, a faster reply to their email, or an early look at a new product. But in a plain Shopify orders list, their order looks exactly like anyone else's. A tag fixes that. Once the customer and the order carry a tag, your team can filter for it, your packing slip can show it, and Shopify's own marketing tools can treat these people differently. Shopify Flow adds the tag for you on every new order, so nobody has to remember. ### How do I know I am missing this? - You find out someone is a regular only when they mention it in an email. - Your team cannot answer "how many customers came back this month?" without an export. - Every customer gets the same order confirmation and nothing else, whether it is their first order or their tenth. If any of those sound familiar, this takes about half an hour and runs on its own after that. Steps: 1. Create a Flow workflow on Order created: Install the free Shopify Flow app if you do not have it, open it, and create a new workflow with the Order created trigger. Staff accounts, unlike the store owner, also need Shopify's Flow automations permission and Edit customers for the tagging step, set in Settings > Users > Roles. 2. Check the customer's order count: Add a condition on the order's customer and pick the number of orders field. Set it to Greater than or equal to 2. 3. Tag the customer and the order: On the true path, add the Add customer tags action with repeat-customer, then add the Add order tags action with repeat-order so the packing team sees it in the Orders list. 4. Turn on the thank-you automation: From your Shopify admin, go to Apps > Messaging, click Automations, then Create automation, and choose Thank customers after they purchase under Post-purchase. Click Edit emails to write both in your own voice: Thank you! for a first order and We're happy to see you again for a second. 5. Send a test, then switch it on: Click Send test and send it to yourself. When both emails read well, go back to the automation and click Turn on automation. 6. Save a segment of repeat customers: In Customers > Segments, create a segment with the query customer_tags CONTAINS 'repeat-customer' so you can see and email this group any time. ### What should I test before I trust it? Place a test order as a brand new customer and check that no tag appears. Then place a second order with the same email and check the customer has repeat-customer and the order has repeat-order. Send the thank-you test email to your own inbox and read it on your phone. If you want the packing team to act on the tag, add a line to your packing slip template or agree a simple rule, such as a handwritten note in every repeat-order box. ### When is this not the right fix? If most of your orders come in as guest checkouts with different emails each time, Shopify cannot always join them to one customer, and the order count will undercount your regulars. If you already run an email platform with its own post-purchase flows, turn on the thank-you there instead, so customers do not get two. Tags only help if someone looks at them, so give one person the job of checking the segment each week. ### What does it cost? Shopify Flow is free. Shopify Messaging includes 10,000 free emails a month on every paid plan, and the thank-you automation counts toward that. Most small stores will not pay anything extra. ### Want this set up for you? We build this, the VIP tier and the packing slip line for free as part of the Founding 10 (/founding-10), where we fix one Shopify problem for ten stores. For a wider look at how your store runs, start with the Store Ops Audit (/store-kits). Questions: - Does the order that triggers the workflow count towards the number? Shopify describes the field as the number of orders the customer has made at the store in their lifetime. Test it with a customer's second order: if the tag does not appear, the count in your store is not including the new order yet, so add a short Wait step before the condition. - Will every customer get the thank-you email? Probably not. Shopify names abandoned checkout as the automation that can be sent to all customers, not just subscribers, so expect the thank-you to reach customers who agreed to email marketing. Check the audience in the automation before you turn it on. - How much does the thank-you email cost? Shopify Messaging gives every store on the Basic, Grow, Advanced and Plus plans 10,000 free emails each month, and automation emails count toward that. After that it is $1 USD per 1,000 extra emails. - Can I do something special for my best customers? Yes. Shopify has a Welcome VIP customers automation that sends when a customer joins the VIP segment, which includes customers tagged VIP. Add a second Flow condition, such as five or more orders, that adds the VIP tag. Sources: - Shopify Help Center: Add customer tags (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/add-customer-tags - Shopify Help Center: Add order tags (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/add-order-tags - Shopify Help Center: Creating marketing automations in Shopify: https://help.shopify.com/en/manual/promoting-marketing/create-marketing/marketing-automations/create - Shopify Help Center: Shopify Email pricing: https://help.shopify.com/en/manual/promoting-marketing/create-marketing/shopify-email/shopify-email-cost - Shopify Help Center: Shopify-based customer segment filters: https://help.shopify.com/en/manual/customers/customer-segmentation/reference-guide/shopify-segments - Shopify Dev: Customer object (numberOfOrders): https://shopify.dev/docs/api/admin-graphql/latest/objects/Customer - Shopify Changelog (28 Oct 2026): Flow staff permissions: https://changelog.shopify.com/posts/shopify-flow-staff-now-need-permissions-to-test-and-turn-on-workflows ### Fix #10: How do I stop overselling during a big sale on Shopify? https://www.axioprax.com/fixes/stop-overselling-shopify-sale (Inventory & reorders, published 2026-11-07) Short answer: Open Products > Inventory, select everything and open the bulk editor with the Track quantity and Continue selling when out of stock columns. Turn tracking on and Continue selling off for every physical variant you cannot restock fast. Then check which locations fulfill online orders and whether every bundle component is tracked. ### Why does overselling happen most during a sale? On a quiet day, a setting that is slightly wrong rarely matters. A variant with tracking switched off, or one left on Continue selling when out of stock, might go a little past zero and you restock before anyone notices. During a sale, the same variant can take dozens of orders in an hour, and every one past your real stock is a customer you have to email, refund and apologise to. Shopify already has everything you need to stop this. By default, Shopify stops selling a tracked product at zero. Overselling comes from the exceptions: products that were never tracked, products someone set to keep selling, stock sitting at the wrong location, and bundles built from parts Shopify cannot count. ### How do I know my store is at risk? - You have cancelled an order before because the item was not on the shelf. - Products show a negative stock number in Products > Inventory. - Products were imported from another platform or app and nobody checked their inventory settings. - You sell bundles or kits made from products you also sell on their own. Any one of these is worth 45 minutes before your next sale. Steps: 1. Open every variant in the bulk editor: From your Shopify admin, go to Products > Inventory, use the checkboxes to select the products and variants, and click Edit variants. 2. Show the two columns that matter: Click Columns and select Track quantity and Continue selling when out of stock, plus Quantities by location if you have more than one location. 3. Turn on Track quantity: Switch Track quantity on for every physical product. Leave it off only for things Shopify suggests not tracking, such as digital products, services, made-to-order items and gift cards. 4. Turn off Continue selling when out of stock: Uncheck it for every variant you cannot restock quickly, keep it only for deliberate pre-orders, then click Save and Save changes. 5. Check which locations sell online: Go to Settings > Locations, open each location, and check the Use inventory at this location to fulfill online orders toggle. Only stock at locations with it on counts toward what online customers can buy. 6. Check bundle components: Open each bundle and confirm every product inside it has Track quantity on and Continue selling when out of stock off, because Shopify ignores components that are untracked or set to continue selling. 7. Test with a quantity of one: Set a test product to 1 in stock, buy it, then try to buy it again from another browser. The second checkout should be blocked. ### What about pre-orders? Continue selling when out of stock is the right setting for a product you are deliberately taking pre-orders for. The problem is when it is switched on by accident, or left on after the pre-order ends. Keep a short list of the variants that should keep selling, and check the bulk editor against that list after every import. ### When is this not the right fix? If you also sell on a marketplace or another platform that keeps its own stock count, Shopify can only stop overselling on the channels it controls. You need those counts to sync, and that depends on the app or connector you use. If your counts are wrong because nobody has done a stock count in months, these settings will faithfully enforce the wrong number, so count the fast sellers first. POS is the other gap. Shopify says the setting does not apply to in-person orders, so brief your shop staff during a sale. ### What does it cost? Nothing extra. Inventory tracking, the bulk editor and location settings are built into the Shopify admin, with no app to install. The cost is about 45 minutes in the bulk editor, plus an honest stock count on the products you expect to sell fastest. ### Want this set up for you? We will audit your inventory settings and fix them before your next sale for free as part of the Founding 10 (/founding-10), where we fix one Shopify problem for ten stores. If you used Stocky for purchase orders, see the Stocky Replacement kit (/store-kits/stocky-rescue). Questions: - Why is a product still selling at zero when Continue selling is off? Check that Track quantity is on. Shopify only lets you choose Continue selling when out of stock once inventory is tracked, and the out-of-stock rules are written for tracked products. - Does this stop my in-store POS staff selling past zero? No. Shopify says the setting does not apply to orders placed in Shopify POS. Staff can still sell past zero, but POS warns them that the item is not available. - Why does a product show out of stock online when I have stock? The stock is probably at a location that does not fulfill online orders. Shopify says you need to add stock to a location that fulfills online orders for it to sell online. - How does Shopify decide how many bundles I can sell? With Shopify's bundles, the component with the lowest inventory sets how many bundles can be sold. Untracked components, and components set to continue selling, are left out of that sum. Sources: - Shopify Help Center: Selling products when out of stock: https://help.shopify.com/en/manual/products/inventory/getting-started-with-inventory/selling-when-out-of-stock - Shopify Help Center: Set up inventory tracking: https://help.shopify.com/en/manual/products/inventory/getting-started-with-inventory/set-up-inventory-tracking - Shopify Help Center: Bulk editing inventory: https://help.shopify.com/en/manual/products/inventory/managing-inventory-quantities/bulk-editing-inventory - Shopify Help Center: Setting up location fulfillments: https://help.shopify.com/en/manual/locations/setting-up-your-location-fulfillments - Shopify Help Center: Eligibility and considerations for product bundles: https://help.shopify.com/en/manual/products/bundles/eligibility-and-considerations ### Fix #11: How do I let customers start a return themselves on Shopify? https://www.axioprax.com/fixes/let-customers-start-returns-shopify (Support, published 2026-11-08) Short answer: Set your return window, return shipping and final sale items in Settings > Policies. Then turn on self-serve returns in Settings > Customer accounts. Customers request a return from their order status page, you get a notification, and you approve or decline it from the order, with a label if you want. ### Why let customers start the return themselves? A return by email usually takes three or four messages: which order, which item, why, and where to send it. Each reply waits in a queue. Meanwhile the customer is unsure whether you have even seen it. Shopify's self-serve returns move that first step to the customer. From their order status page they pick the items and send a request, Shopify checks it against the rules you set, and you get a notification. The request lands on the order itself, so you decide with all the details in front of you, and the customer gets an email either way. ### How do I know I need this? - Your support inbox has more return emails than product questions. - Customers ask for returns after your window has closed, or on items you never take back. - You cannot quickly answer "how many returns are waiting right now?" Before you switch it on, decide your rules. Writing the policy first is what makes self-serve work, because Shopify uses those rules to decide which items a customer can even request. Steps: 1. Set your return rules: From your Shopify admin, go to Settings > Policies and open your rule in the Return and cancellation rules section. Choose the return window, the return shipping option and any restocking fee, then save. 2. Mark final sale items: In the same rule, add final sale items by choosing Collections or Products under Applies to. Customers cannot return or cancel these items. 3. Check customer accounts: Self-serve returns do not work with legacy customer accounts. In Settings > Customer accounts, make sure your store uses the newer customer accounts. 4. Turn on self-serve returns: In Settings > Customer accounts, if you see Order actions, click Manage and turn on Return. If not, turn on Self-serve returns and cancellations and choose the request type. 5. Edit the customer emails: Go to Settings > Notifications > Returns and read Return request confirmation, Return request approved and Return request declined, so they sound like your store. 6. Review requests from one filter: In Orders, use the Customer request filter. Open an order, click Review request, choose a return shipping option and click Approve, or click Decline and pick a reason. ### What should I test before I trust it? Place a test order, fulfil it, then sign in as that customer and request a return. Check the confirmation email arrives, the order shows under the Customer request filter, and the Return requested card appears. Approve it once with a label and once with No shipping required, so you know what each email looks like. Then try to return a final sale item and confirm the customer cannot. ### When is this not the right fix? If you sell mostly custom or made-to-order items that you never take back, a clear policy page matters more than a request button. If you already use a returns app with its own portal, keep one system, or customers will find two ways in. And remember the gaps: returns only work on fulfilled items, return fees are not taken off refunds for you, and rule changes only affect new orders. ### What does it cost? Self-serve returns and return rules are built into the Shopify admin. Labels you create or upload cost whatever your carrier charges, and you choose whether the customer pays through the return shipping setting. ### Want this set up for you? We set up your rules, the customer emails and a daily review routine for free as part of the Founding 10 (/founding-10), where we fix one Shopify problem for ten stores. For a wider look at your support load, start with the Store Ops Audit (/store-kits). Questions: - Can Shopify create the return label for me? Only in some cases. Shopify can create a return label if you have a location in the same country as the customer and that country is the United States or Canada. Otherwise you can upload a label or choose No shipping required. - Does approving a return refund the customer? No. Approving sends the customer the return details. You refund and restock after the items come back and you have inspected them. - Are restocking and return shipping fees taken off the refund automatically? No. Shopify says return fees are not deducted from refunds automatically, so you deduct them yourself when you refund. - Do new return rules apply to old orders? No. Changes to return rules apply only to future orders, so orders placed before you change the rule keep the old terms. Sources: - Shopify Help Center: Self-serve returns and cancellations: https://help.shopify.com/en/manual/orders/self-serve-returns - Shopify Help Center: Setting up self-serve returns: https://help.shopify.com/en/manual/orders/return-requests - Shopify Help Center: Setting up return and cancellation rules: https://help.shopify.com/en/manual/orders/self-serve-returns/return-rules - Shopify Help Center: Managing self-serve returns: https://help.shopify.com/en/manual/orders/self-serve-returns/managing-returns ### Fix #12: What should I do when a customer opens a chargeback on Shopify? https://www.axioprax.com/fixes/respond-to-chargeback-shopify (Fraud & bots, published 2026-11-08) Short answer: Open the order in Orders, read the chargeback banner and note the due date, usually 7 to 21 days after it was filed. Do not try to refund. Gather evidence that matches the reason, such as tracking and delivery proof for item not received, click Add evidence, put the strongest proof first, and submit before the deadline. ### What actually happens when a customer opens a chargeback? The customer tells their bank they want their money back. With Shopify Payments, the bank takes the disputed amount from you straight away, plus a chargeback fee. You then have a window to send evidence, and the card company decides. An inquiry is the softer version: the bank asks questions first and does not take the money or a fee right away. The part that catches stores out is timing. Shopify gives the due date as 7 to 21 days after the chargeback is filed, and once you submit, you cannot add anything. After the bank decides, Shopify says the decision is final. ### How do I know I have one? - An email to your store contact email address. Check which address that is in Settings > Store details. - A chargeback banner at the top of the order. - The order showing under the Chargeback and inquiry status filter in Orders. If the store contact email goes to an inbox nobody reads, fix that first. A missed email is the easiest way to lose by default. Steps: 1. Find every open chargeback: From your Shopify admin, go to Orders, click Search and filter, add the Chargeback and inquiry status filter and select Open. 2. Read the reason and the due date: Open the order and read the chargeback banner. Write down the reason and the due date, which Shopify shows in your store's timezone. 3. Check whether Shopify Protect covers it: If the order used Shop Pay and was protected by Shopify Protect, and the reason is fraudulent or unrecognized, Shopify says you do not need to take any action. 4. Contact the customer once: Email the customer politely. If it was a mistake, ask them to withdraw the chargeback with their bank and send you the bank's withdrawal letter. 5. Gather evidence that matches the reason: Collect proof for that specific reason, such as tracking and delivery confirmation, your product page as it looked at purchase, your policies and every customer message. Save each as a PDF, JPEG or PNG under 2 MB. 6. Add evidence and submit: In the chargeback banner, click Add evidence, add your summary, upload the files and click Save. Click Submit now when you are done, or Shopify submits it on the due date. ### What evidence works best? Match the evidence to the reason. For item not received, Shopify lists tracking information, delivery confirmation ideally with a signature or photo, proof it went to the checkout address, and carrier notifications. For fraudulent, it lists AVS and CVV matches, past successful orders from the same customer, and delivery to the verified address. Shopify's advice is to put your strongest proof first, label every file clearly, and use screenshots of tracking pages and messages rather than links. Shopify also says, in its 28 October 2026 Shopify Protect announcement, that new AI tooling now gathers card-network evidence, pulls delivery photos from carriers and drafts dispute responses for Shopify Payments stores. If you see a drafted response, check it and add your own proof, such as photos of the packed box, before you submit. ### Will this win the chargeback? We cannot promise that, and neither does Shopify. The card company decides, and a well-built response simply gives you your best chance. Sometimes accepting is the right call, for example when the customer has a fair point and you have no delivery proof. You accept by not submitting evidence, and you do not get the fee back. ### When is this not the right fix? This guide is for Shopify Payments. If you use another payment provider, disputes are handled in that provider's dashboard with its own deadlines. And if chargebacks keep coming from the same people, responding one by one is not enough. Block them at checkout instead. ### What does it cost? Responding is free. The bank takes a chargeback fee when the case opens. If you win, the amount comes back, and the fee may too depending on your country or region. The real cost is the hour it takes to build a clean evidence file. ### Want this set up for you? Our Fraud Hold kit (/store-kits/fraud-hold) holds risky orders before they ship, so fewer turn into chargebacks. We can also set up an evidence-ready order process for free as part of the Founding 10 (/founding-10). Questions: - Can I just refund the customer to make the chargeback go away? No. Shopify says you cannot issue a refund after a cardholder starts a chargeback. The customer has to withdraw it with their bank first. For an inquiry, a full refund ends it, but a partial refund can still lead to a full chargeback. - Do I get the chargeback fee back? If you win, the amount comes back in your next payout after the decision, and Shopify says the fee may be refunded too depending on your country or region. If you accept the chargeback, you do not get the fee back. - Can I add more evidence after I submit? No. Once evidence is submitted you cannot change it or add more, so only click Submit now when everything is in. - Does Shopify Protect cover every chargeback? No. It covers fraudulent or unrecognized chargebacks on protected orders that used Shop Pay. Other reasons, such as item not received or not as described, follow the normal process. Sources: - Shopify Help Center: Chargebacks and inquiries: https://help.shopify.com/en/manual/payments/chargebacks - Shopify Help Center: Managing chargebacks in the Shopify admin: https://help.shopify.com/en/manual/payments/chargebacks/chargebacks-in-admin - Shopify Help Center: Responding to chargebacks and inquiries: https://help.shopify.com/en/manual/payments/chargebacks/chargeback-process - Shopify Help Center: Resolving a chargeback or inquiry: https://help.shopify.com/en/manual/payments/chargebacks/resolve-chargeback - Shopify Help Center: Addressing chargebacks with Shopify Protect: https://help.shopify.com/en/manual/payments/shop-pay/shopify-protect/addressing-chargebacks - Shopify blog: Introducing Shopify Protect (28 Oct 2026): https://www.shopify.com/blog/introducing-shopify-protect ### Fix #13: How do I test my Shopify checkout without real money? https://www.axioprax.com/fixes/test-shopify-checkout-without-real-money (Checkout & orders, published 2026-11-09) Short answer: Go to Settings > Payments, click Manage under Shopify Payments and check Enable test mode. Place an order with test card 4242 4242 4242 4242, any future expiry and any three digits, then check the order, the emails, tax, shipping and stock. Turn test mode off straight after, because real customers cannot pay while it is on. ### Why test checkout if nothing looks broken? Checkout is where small changes show up. A new shipping rate, a tax setting, a discount, a payment app or a theme update can each change what a customer pays or sees, and the first person to notice is usually a customer who gives up. A test order shows you exactly what they get, from the cart to the confirmation email. Shopify recommends placing at least one test order during setup and whenever you change your payment settings. Its own list of what a test order checks is checkout, order processing, inventory, shipping, email notifications and taxes. That is most of what goes wrong. ### How do I know I should run one now? - You changed payments, shipping rates, taxes, markets or discounts since your last test. - You installed or removed an app that touches checkout, orders or fulfillment. - A customer told you checkout did something odd and you could not reproduce it. If you cannot remember your last test order, run one this week. Steps: 1. Pick a quiet time and pause auto-fulfillment: Shopify advises against test mode on a live store, because real customers cannot pay while it is on. If you use it on a live store, keep it to a few minutes at your quietest hour, or place one real order and refund it instead. Either way, pause any fulfillment service or auto-fulfillment app first. 2. Turn on test mode: From your Shopify admin, go to Settings > Payments. Under Shopify Payments, click Manage, check Enable test mode in the Test mode section and save. 3. Place a successful test order: Check out as a customer with an email ending in @example.com and card 4242 4242 4242 4242, a name of at least two words, any future expiry date and any three digits. The order must be over the equivalent of $1 USD. 4. Place a declined test order: Check out again with 4000 0000 0000 0002, Shopify's generic decline card, and read the error message your customer would see. 5. Check what the order produced: Open the order and confirm the test mode banner, then check tax, shipping, discounts, stock and the order confirmation email. Add a tracking number to check the shipping email, but do not buy a label. 6. Turn test mode off and clean up: Uncheck Enable test mode and save straight away, then delete the test orders and switch auto-fulfillment back on. ### What should I check on the test order? - The total: product price, discount, shipping and tax all match what you expect for that address. - Stock: the quantity went down for a tracked product. - Emails: the order confirmation, and the shipping confirmation once you add tracking. - Your apps: anything that tags, routes or notifies on new orders did its job. Test one address in each country or region you ship to most. Shipping and tax both change by address, so a single test only proves one route. ### When is this not the right fix? If you take payments through express buttons, be careful. Shopify says Apple Pay and Google Pay use real cards in test mode, though no real charges are processed, and PayPal Wallet does not show at all. If those are most of your payments, also run one real order and refund it. Test mode also needs a paid plan, finished Shopify Payments setup and two-step authentication. ### What does it cost? Nothing, as long as you do not fulfil the test orders. Shopify warns you are charged for any shipping labels you buy. A real-card test costs whatever fees your provider keeps when you refund. ### Want this set up for you? We run a full checkout test across your main markets and write up what we find for free as part of the Founding 10 (/founding-10), where we fix one Shopify problem for ten stores. For a broader check of how your store runs, start with the Store Ops Audit (/store-kits). Questions: - Is it safe to turn on test mode on a live store? Only briefly. Shopify says you should not activate test mode on a store that is in production, and real customers cannot place orders while it is on. Keep the window short and do it at your quietest hour. - What can I not test in test mode? Shopify says test mode does not cover POS card readers, Shop Pay Installments or PayPal Wallet. Manual payment methods such as cash on delivery do not support test mode either, so test those with a real order. - What if I do not use Shopify Payments? Use your provider's own test mode, or place a real order and immediately cancel and refund it. Shopify notes some provider and transaction fees are not refunded. - Can I delete test orders afterwards? Yes, if they were paid with Shopify Payments in test mode or the Test payment gateway. Orders through other gateways are not counted as test orders and cannot be deleted, but you can archive them. Sources: - Shopify Help Center: Testing Shopify Payments: https://help.shopify.com/en/manual/payments/shopify-payments/testing-shopify-payments - Shopify Help Center: Placing a test order: https://help.shopify.com/en/manual/checkout-settings/test-orders - Shopify Help Center: Processing a test order: https://help.shopify.com/en/manual/checkout-settings/test-orders/processing-test-order ### Fix #14: How do I hide sold-out products on Shopify without deleting them? https://www.axioprax.com/fixes/hide-sold-out-products-shopify (Inventory & reorders, published 2026-11-10) Short answer: To take sold-out products out of collections but keep their pages live, add the condition Inventory stock is greater than 0 with Match all conditions to each automated collection. To hide them from the online store completely and bring them back on restock, use the free Shopify Flow template Publish and unpublish products based on inventory level. ### Why not just delete sold-out products? Deleting a product removes its page, its reviews, its sales history and any link pointing to it. When the stock comes back, you rebuild it from scratch. Leaving it visible is not great either: shoppers scroll past items they cannot buy, and some of them leave. Shopify has two free in-between options. An automated collection can quietly drop products at zero stock and add them back on restock, while the product page itself stays live. Or Shopify Flow can unpublish the product from the online store at zero and publish it again when stock returns. Both undo themselves, so nobody has to remember. ### How do I know I need this? - Your collection pages open with a row of Sold out badges. - Someone on your team hides and unhides products by hand after each delivery. - You have deleted products in the past and later recreated them. Before you choose, ask one question: do people still land on these product pages from Google, ads or emails? Your answer decides which option fits. Steps: 1. Make sure inventory is tracked: Both methods only work for products with Track quantity on. Check this in Products > Inventory with the bulk editor before you start. 2. Option A: add the stock condition to each collection: Open each automated collection, change the collection conditions to add Inventory stock is greater than 0, and set products to match all conditions. Save. 3. Replace the default Catalog page if you use it: The built-in collection at /collections/all shows every product. Override it with your own automated collection that has the same stock condition. 4. Option B: install the Flow template: Open the free Shopify Flow app, search the templates for Publish and unpublish products based on inventory level, and install it. Staff accounts, unlike the store owner, also need Shopify's Flow automations permission, set in Settings > Users > Roles. 5. Set which channels it hides from: In the Unpublish product and Publish product actions, choose the publications, usually your Online Store, then turn the workflow on. 6. Test both directions: Set a test product to 0 and check it disappears, then set it back to 1 or more and check it returns. Shopify's own test advice for the out of stock trigger is to change a variant from above 0 to 0. ### What is the difference for my product page? With the collection condition, the product is only missing from collections. Its own page still works, so a shopper arriving from search can read about it, and you can offer a back in stock email there. With Flow, the Unpublish product action makes the product unavailable on the sales channels you choose. That is cleaner for browsing, but anyone following an old link reaches a page that is no longer available until the product is published again. ### When is this not the right fix? If a product is discontinued for good, hiding it is just a slower delete. Decide whether to keep the page for search or archive it. If your stock numbers are not reliable, these methods will hide products you actually have, so fix tracking first. And if you use manual collections, the stock condition is not available there, so Flow is your only automatic option. ### What does it cost? Nothing. Automated collections are built into Shopify, and Shopify Flow is a free app on the Basic, Grow, Advanced and Plus plans. Setup takes about half an hour, most of it opening each collection. ### Want this set up for you? We set this up and test it on your collections for free as part of the Founding 10 (/founding-10), where we fix one Shopify problem for ten stores. If you used Stocky to manage reorders, see the Stocky Replacement kit (/store-kits/stocky-rescue). Questions: - Which option should I choose? Use the collection condition if sold-out products get traffic from Google, ads or emails, because the product page stays live. Use the Flow template if you want the product gone from the online store until it is back in stock. - Why is a sold-out product still showing in my collection? Check two things. Shopify says the condition does not work if the collection is set to match any condition, and products need inventory tracking turned on. - When does the Flow template bring a product back? Shopify describes the template as republishing products once they have at least one variant with inventory. The back in stock trigger fires when a variant goes from 0 or less to 1 or more. - Will this hide my pre-order products? It can, if they sit at 0 stock. Exclude them with a tag in your Flow condition, or keep pre-orders in a manual collection without the stock condition. Sources: - Shopify Help Center: Hiding out of stock products: https://help.shopify.com/en/manual/products/inventory/product-availability - Shopify Help Center: Unpublish product (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/hide-product - Shopify Help Center: Publish product (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/publish-product - Shopify Help Center: Product variant out of stock (Flow trigger): https://help.shopify.com/en/manual/shopify-flow/reference/triggers/variant-out-of-stock - Shopify Help Center: Product variant back in stock (Flow trigger): https://help.shopify.com/en/manual/shopify-flow/reference/triggers/variant-back-in-stock - Shopify Changelog (28 Oct 2026): Flow staff permissions: https://changelog.shopify.com/posts/shopify-flow-staff-now-need-permissions-to-test-and-turn-on-workflows ### Fix #15: Can I set up a backup payment method in case Shopify Payments stops? https://www.axioprax.com/fixes/backup-payment-method-shopify (Payouts & holds, published 2026-11-11) Short answer: Partly. Outside the US and France, finish setting up PayPal in Settings > Payments so shoppers can still pay with PayPal. PayPal does not replace a card provider, so for cards you need a direct third-party provider, either alongside Shopify Payments or switched to. Test both before Black Friday. None of this releases held funds. ### What does a backup payment method actually cover? Shopify splits payments into two kinds. A credit card provider, such as Shopify Payments, takes card payments. Additional payment methods sit next to it. Shopify is clear on this: "Additional payment methods like PayPal process only PayPal payments and don't replace a primary provider." So if Shopify Payments stops and PayPal is your only backup, shoppers who want to pay by card cannot. Shopify says they may see "This store is currently unable to accept payments", which usually means no active primary payment provider (the one that takes cards) is set up. A real backup covers both: PayPal for PayPal buyers, and a second card provider for everyone else. Be honest with yourself about the limits. A backup does not release money Shopify Payments is already holding, and it does not stop Shopify from reviewing or pausing your account. If a payout is already on hold, start with Fix #2 (/fixes/shopify-payout-on-hold-what-to-prepare) instead. ### How do I know I need this before Black Friday? - PayPal shows Setup incomplete in Settings > Payments. - Shopify Payments is your only way to take cards, and nobody has the login for another provider. - You have never placed a test order through anything except Shopify Payments. - Your best week of the year is close, and a day without card payments would hurt. If two of these are true, give this an hour this week, while nothing is wrong. Steps: 1. Check your country first: In the United States and France, PayPal is integrated directly with Shopify Payments and is not available as a separate payment gateway, so it is not an independent backup there. Everywhere else, carry on with step 2. 2. Finish your PayPal setup: Go to Settings > Payments. In the Additional payment methods section, click Setup incomplete next to PayPal, then click Complete setup. Shopify says refunds and manual capture need this setup to be complete. 3. Make a real PayPal test purchase: Create a test product and buy it from your store with a different PayPal account from the one that receives your payments. Check the payment arrives in PayPal. 4. Decide how you will take card payments: PayPal takes only PayPal payments. For cards, you can run one direct third-party provider alongside Shopify Payments, or keep the provider's credentials ready to switch to in Settings > Payments. 5. Know the switch path before you need it: In Settings > Payments, Shopify payments section, click Manage, then click Switch to a third-party provider at the bottom of the page and click it again to confirm. Choose the provider, enter its credentials, then click Activate and Save. This removes Shopify Payments as your card provider, and you can have only one credit card provider activated at a time this way. 6. Test any new card provider: Use the provider's test mode if it offers one in your Shopify admin, and place a test order. If it does not, ask the provider's support team how to test it with Shopify. Do not switch to Shopify's Test payment gateway on a live store, because you have to deactivate your current card provider first. Turn test mode off when you finish, because real cards do not work while it is on. 7. Watch the banner on your admin homepage: Shopify says that if your Shopify Payments account is on hold, you should review the banner message on the homepage of your Shopify admin and check your email. ### How does a second card provider alongside Shopify Payments work? Shopify lets stores on all plans except contracted Shopify Plus use one direct third-party provider alongside Shopify Payments. Plus stores with a contract can request it, and multi-entity stores cannot use it. In this setup the third-party provider takes card payments first, and Shopify Payments retries eligible declined card payments as a backup. Wallets and local payment methods are not retried. One catch: Shopify's page describes it from the other side, a store already on a third-party provider activating Shopify Payments next to it. If you are on Shopify Payments today, ask Shopify Support how to add the provider without switching away. And Shopify's pages do not say what happens to this setup if Shopify Payments itself is paused. ### When is this not the right fix? If your payout is already on hold, a backup will not get that money back. Use Fix #2 to prepare for the review. If you are in the United States or France, PayPal is part of Shopify Payments, so it does not count as a backup. And if Shopify Payments is your only card option in your country, focus on keeping that account in good standing. ### What does it cost? A third-party card provider charges its own fees, and Shopify adds a third-party transaction fee on transactions through third-party and alternate gateways, even when Shopify Payments is active. PayPal transactions are exempt from that fee only while Shopify Payments is your primary provider, so check the rate on your plan before you switch. ### Want this set up for you? We set this up and test it with you for free as part of the Founding 10 (/founding-10), where we fix one Shopify problem for ten stores. To check the rest of your checkout first, see Fix #13 (/fixes/test-shopify-checkout-without-real-money). Questions: - Will adding PayPal release my held payouts? No. A backup only gives shoppers another way to pay from now on. It does not change money Shopify Payments is already holding. For that, see Fix #2. - If Shopify Payments is turned off, does PayPal keep working? In the United States and France, PayPal runs inside Shopify Payments, so it is not a separate backup. Elsewhere, Shopify's pages do not say what happens in that case, so ask Shopify Support before you rely on it. - Can I have two card providers at once? On most plans you can use one direct third-party provider alongside Shopify Payments, but not two. Shopify Plus stores with a contract need to request access, and multi-entity stores cannot use it. - What do shoppers see if no card provider is working? Shopify says they may see This store is currently unable to accept payments. That usually means no active primary payment provider (the one that takes cards) is set up. Sources: - Shopify Help Center: Setting up PayPal: https://help.shopify.com/en/manual/payments/paypal/set-up-paypal - Shopify Help Center: Troubleshooting payments: https://help.shopify.com/en/manual/payments/troubleshooting - Shopify Help Center: Activating a credit card payment provider: https://help.shopify.com/en/manual/payments/third-party-providers/configuring-providers - Shopify Help Center: Using Shopify Payments with a third-party provider: https://help.shopify.com/en/manual/payments/shopify-payments/using-shopify-payments-with-a-third-party-provider - Shopify Help Center: Third-party payment providers: https://help.shopify.com/en/manual/payments/third-party-providers - Shopify Help Center: Payments test mode: https://help.shopify.com/en/manual/checkout-settings/test-orders/payments-test-mode ### Fix #16: How do I turn off Shopify checkout inside Google AI Mode? https://www.axioprax.com/fixes/turn-off-google-ai-mode-checkout-shopify (Checkout & orders, published 2026-11-12) Short answer: Go to Sales channels > Agentic in your Shopify admin, turn off Allow Shopify to manage for me, click Google AI Mode and Gemini, turn off Direct checkout and click Save. Your products can still be found there, but shoppers are sent to your online store to finish the purchase. ### What is direct checkout in Google AI Mode? Shopify's agentic storefronts put your products in front of shoppers inside AI chat tools, and Google AI Mode and Gemini are two of them. With direct checkout on, Shopify says customers can complete their purchase in the channel's Shopify-powered direct checkout, so they never land on your store to pay. Many owners never chose this. Shopify says agentic storefronts are active by default for eligible stores, and the default setting, Allow Shopify to manage for me, keeps all channels active, turns on direct checkout where a channel offers it, and enrols you in new AI channels as they launch. ### How do I know this applies to my store? - You sell to customers in the US, Canada or Australia. Shopify's Google help page lists this as a requirement, and says your store can be based in any country. - Your products already reach Google. Shopify says they get to Google AI Mode and Gemini through Google Merchant Center, usually by way of the Google & YouTube sales channel. - Orders show up attributed to an AI channel you do not remember setting up. Shopify says each order is attributed to the AI channel that drove the sale. The quickest check is the admin itself. Shopify says direct checkout for Google is still rolling out and might not be in your store yet, and the Agentic section tells you whether you are eligible. Steps: 1. Open the Agentic section: In your Shopify admin, go to Sales channels > Agentic. It shows which channels offer direct checkout and whether your store is eligible. You need to be the store owner, or a staff member with the App development or App and sales channel permission for Agentic. 2. Turn off Allow Shopify to manage for me: This is the default setting, and while it is on, Shopify keeps every channel active and switches direct checkout on wherever a channel supports it. Turn it off so you can change the settings underneath it. 3. Leave Shopify Catalog access on: Catalog access is how AI channels get your products. Leave it on if you still want to be found, because turning it off also turns off direct checkout and stops sharing your product data with third-party agentic storefronts. 4. Turn off Direct checkout for Google AI Mode and Gemini: Click Google AI Mode and Gemini, then deactivate Direct checkout. Shoppers can still find your products there, but they get redirected to your online store to finish the purchase. 5. Decide on Auto enroll new storefronts: If you want to approve every new AI channel yourself, turn off Auto enroll new storefronts. You can still enrol in a channel by hand later. Then click Save. 6. Do not switch the managed setting back on: Shopify says turning Allow Shopify to manage for me back on reactivates all agentic storefront settings, including direct checkout. Write a note for your team so nobody flips it back during a clean-up. ### When should I leave it on? Often, honestly. Direct checkout removes a step: the shopper pays without leaving the conversation. Shopify says you stay the merchant of record, keep the customer relationship, see each order in your admin and pay no transaction fees beyond standard processing rates. If you sell simple products to eligible shoppers, there is a fair case for leaving it alone. Turn it off when your own checkout does work you rely on. Shopify notes that some checkout blocks, such as those for custom fields, upsells or specific validations, might not display in Google's direct checkout. If you depend on them, sending shoppers to your store keeps them in place. ### When is this not the right fix? If you want your products out of AI channels completely, turning off checkout is not enough. Turning off Shopify Catalog access stops third-party agentic storefronts reaching your products through Catalog, although Shopify says it can take up to 7 days and products can still be found through web crawling. Shop has no Catalog access switch in the Agentic section. Shopify says it is managed in the Shop sales channel settings instead. The honest limit: none of these switches guarantee your products never appear in an AI answer. ### What does it cost? Nothing. These are free settings in your Shopify admin, and changing them adds no fees. Allow time to read what each switch does before you save. Questions: - Is direct checkout in Google AI Mode really on by default? For eligible stores, yes. Shopify says Allow Shopify to manage for me is the default, and with it on, direct checkout is activated for the channels that offer it. Stores on the Agentic plan are different: direct checkout starts off there. - Will my products disappear from Google AI Mode if I turn checkout off? No. Turning off Direct checkout only changes where the purchase happens. Shoppers who want to buy are sent to your online store instead. - Which stores can use it? Shopify says Google AI Mode and Gemini is rolling out to eligible stores that sell to customers in the United States, Canada or Australia, and your store can be based in any country. Direct checkout is rolling out gradually to shoppers in the United States, Canada and Australia, and shoppers who do not have it yet are sent to your online store to pay. The Agentic section in your admin shows whether you are eligible. - How do I hide a product from every AI channel? Set its status to Unlisted. Shopify warns this also hides it from sitemaps, search engines such as Google and your online store search, so use it only for products you really want out of sight. Sources: - Shopify Help Center: Selling on Google AI Mode and Gemini: https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/google - Shopify Help Center: Managing agentic storefronts: https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/agentic-home - Shopify Help Center: Agentic storefronts: https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts - Shopify Help Center: Shopify Catalog and product discovery: https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/products - Shopify Help Center: Customer experience on agentic storefronts: https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/customer-experience - Shopify blog: How agentic commerce works: https://www.shopify.com/blog/how-agentic-commerce-works ### Fix #17: How do I auto-hold high-risk orders before they ship? https://www.axioprax.com/fixes/auto-hold-high-risk-orders-shopify (Fraud & bots, published 2026-11-13) Short answer: Build a Shopify Flow workflow that starts with the Order risk analyzed trigger, checks that the risk level is high, then runs the Hold fulfillment order action, Add order tags and Update order note. Held orders cannot be fulfilled. Staff review the fraud analysis, then click Release hold or cancel the order. ### What is a fulfillment hold, and why use one for risky orders? Shopify puts it simply: "If there's an issue that prevents you from fulfilling an order, then you can mark your order as on hold." While an order is on hold, its stock can stay reserved, but it cannot be shipped until someone releases the hold. High fraud risk is one of the reasons Shopify gives for using it. That matters because the expensive part of a fraudulent order is usually the parcel. Once it leaves, you have lost the goods and the postage, and a chargeback may follow anyway. A hold buys you time to look before anything is packed, without cancelling an order that might turn out to be a real customer. You can place holds by hand, but on a busy day nobody checks every order's risk level. Shopify Flow, which Shopify describes as a free ecommerce automation platform, can do it for you as soon as the fraud analysis is done. ### How do I know I need this? - You have shipped an order that later turned into a fraud chargeback. - Your team packs orders straight from the Unfulfilled list without opening each one. - You use automatic payment capture and do not want to change it, but still want a check before shipping. - You find out an order was high risk only after it has gone. Steps: 1. Look for the template first: In your Shopify admin, go to Apps > Flow, click Create workflow, then Browse templates and search for Hold fulfillments based on risk level. If it is not in your library, start from a blank workflow and follow the next steps. 2. Start with Order risk analyzed: Choose the Order risk analyzed trigger, not Order created. It starts the workflow when the order's fraud analysis is complete, so the risk level actually exists when Flow checks it. 3. Add a high-risk condition: Add a condition on the order's risk assessment risk level being High (in Shopify's Admin API this is OrderRiskAssessment.riskLevel with the value HIGH). Do not use the older order risk level field, which Shopify has deprecated. Everything else ships as normal. 4. Hold the fulfillment: On the true branch, add the Hold fulfillment order action. Pick a Reason, write Reason notes such as Check fraud analysis before release, and select Notify merchant. 5. Tag and note the order: Add the Add order tags action with one tag, for example risk-review, then Update order note. Start the note text with {{order.note}} so any note the customer left stays on the order, then add your own line. 6. Review, then release or cancel: Open the order from Orders, read its fraud analysis, and in the On hold section click Release hold if it looks genuine. If it does not, cancel the order instead. 7. Test, then turn it on: Turn the workflow on, then place a test order. Shopify says the test order has to trigger its fraud analysis. Check the workflow's run history to see which branch it took, and if it was scored high risk, confirm the hold, tag and note appear, then release the hold. ### How should we review held orders each day? Held orders drop out of the Unfulfilled list, which is good for packing and bad for remembering. Pick one person and one time of day. They search orders for the review tag, open each one, read the fraud analysis, and make a decision. If the order looks genuine, release the hold and it returns to Unfulfilled. If it does not, cancel it. Shopify Flow also has a Release fulfillment order holds action and a Cancel order action, but we suggest a human makes this call until you trust the pattern. ### When is this not the right fix? The honest limit: the Order risk analyzed trigger only runs for Shopify's own risk assessments, not third-party fraud apps. If you rely on another provider's scores, you need that app's own Flow trigger or settings. A hold also does nothing about payment. If you want to avoid capturing money on risky orders as well, pair this with manual capture as in our card-testing fix. And if you sell instant digital downloads, a fulfillment hold may not help, because many download apps deliver the file as soon as the order is paid. ### What does it cost? Nothing extra. Shopify describes Flow as a free ecommerce automation platform. The real cost is a few minutes a day reviewing held orders, plus the occasional delay for a genuine customer whose order was flagged. ### Want this set up for you? We set this up and test it on your store as part of the Founding 10 (/founding-10). If you want a named reviewer and a full hold routine, see the Fraud Hold kit (/store-kits/fraud-hold). Questions: - Is there a ready-made Flow template for this? Shopify's Hold fulfillment order page lists an example called Hold fulfillments based on risk level, which holds the order when the risk level is high. It does not appear in every template list Shopify publishes, so if you cannot find it, build the same thing from a blank workflow: one trigger, one condition and three actions. - Why not use the Order created trigger? Order created can fire before Shopify has finished its fraud analysis. Order risk analyzed waits for the analysis, which is why Shopify notes these workflows do not run immediately after an order is placed. - Does the hold stop the payment being taken? No. A fulfillment hold stops the order being shipped, it does not change how payment is captured. If you also want to delay capture, see our card-testing fix, which uses manual capture. - Where do held orders go in my order list? Shopify says on-hold orders do not appear in the Unfulfilled list. Filter or search by your review tag so held orders get looked at every day. Sources: - Shopify Help Center: Holding fulfillments: https://help.shopify.com/en/manual/orders/fulfillment/fulfilling-orders/holds - Shopify Help Center: Hold fulfillment order (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/hold-fulfillment - Shopify Help Center: Order risk analyzed (Flow trigger): https://help.shopify.com/en/manual/shopify-flow/reference/triggers/order-risk-analyzed - Shopify Help Center: Managing high-risk orders with Shopify Flow: https://help.shopify.com/en/manual/orders/flow - Shopify Help Center: Update order note (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/update-order-note - Shopify Dev Docs: OrderRiskAssessment (risk level field): https://shopify.dev/docs/api/admin-graphql/latest/objects/OrderRiskAssessment ### Fix #18: Why did my Shopify conversion rate suddenly jump? https://www.axioprax.com/fixes/shopify-conversion-rate-jump-bot-sessions (Fraud & bots, published 2026-11-14) Short answer: Most likely Shopify's session measurement update, which rolled out from September 21 to 23, 2026. It filters identified bot sessions out of session-related reports by default, so sessions can drop and conversion rate can rise. Orders and sales are not affected, so check them for the same dates before you celebrate. ### What changed in Shopify's session numbers? In September 2026 Shopify changed how Analytics measures sessions. The rollout ran from September 21 to 23, 2026. Once it reached your store, identified bot sessions were filtered out of session-related reports by default. Sessions are also now based on continued customer activity, instead of ending at midnight. Conversion rate is a share of sessions. If bots made up a real part of your sessions, removing them leaves fewer counted sessions for the same orders, so the rate goes up. Shopify lists sessions, conversion rate, add to cart rate, reached checkout rate, bounce rate and online store visitors among the metrics that may look different. In Shopify's words: "This doesn't necessarily mean that your traffic, orders, or customer behavior changed." ### How do I know this is what happened? - The jump started between September 21 and 23, 2026, with no campaign or price change to explain it. - Sessions fell at the same time, while orders and sales stayed about where they were. - In your sessions reports, the Human or bot session filter now leaves out bot sessions by default. If all three are true, you are looking at a counting change. The steps below help you confirm it and set up a fair comparison going forward. Steps: 1. Match the jump to the update dates: Note the day your conversion rate moved. Shopify's session measurement rollout ran from September 21 to 23, 2026, so a jump in that window points to the update, not your store. 2. Open a sessions report: In your Shopify admin, go to Analytics > Reports and click a sessions report. If the configuration panel is hidden, click Controls. 3. Check the Human or bot session filter: In the Filters section, find the Human or bot session row. Choose is Human to see only human sessions or is Bot to see only bot sessions. To see the full mix, choose is one of, select Human and Bot, and click Apply. 4. Split the report by session type: To see human and bot sessions side by side, add Human or bot session from the Dimensions menu in the configuration panel. Filter changes apply only to the report you are viewing. 5. Check how far back the filter goes: Bot filtering applies only to data from October 7, 2025 onward. Shopify cannot classify older sessions, so reports that reach back before that date still include bots. 6. Compare only periods after the update: Treat September 21, 2026 as a new baseline, as Shopify suggests. Because the rollout ran until September 23, compare periods that start after September 23 with each other, and be careful comparing session metrics with earlier periods. 7. Check orders and sales for the same dates: Open your sales and orders reports in Analytics > Reports for the same date range. These counts are not changed by the update, so they show whether real performance moved. ### When is this not the right fix? If your orders and sales rose too, something real may have changed, and the update only explains part of the jump. Look at your campaigns, prices and traffic sources for the same dates. If the jump happened outside late September 2026, this update is not the cause. Shopify also consolidated marketing reports around the same time, so marketing reports for late September 2026 can reflect both changes. One honest limit: Shopify's bot classification is conservative, so some bots can still be counted as human, and a sudden bot wave can still push sessions around. ### What does it cost? Nothing. The filter is built into Shopify Analytics. Checking it means opening a few reports and lining up the same date ranges. ### What should I tell my team or my agency? Tell them September 21, 2026 is a new baseline. Shopify says to treat it that way when comparing reports before and after the update. Any goal or bonus tied to conversion rate should be reset against post-update numbers. Orders and sales did not change because of this update, so they still compare cleanly with older periods. ### Want help with bot traffic? If bots are hitting your store hard enough to matter, see the Bot Cleanup kit (/store-kits/bot-cleanup). We also fix one Shopify problem for free for ten stores as part of the Founding 10 (/founding-10). Questions: - Did my store actually get better? Not necessarily. Shopify says the change doesn't necessarily mean your traffic, orders or customer behavior changed. Your order and sales numbers are the honest test. - Can I see the bot traffic again? Yes, in reports that support it. Change the Human or bot session filter to include Bot, or remove the filter to show all sessions. - Why can I not filter bots in older data? Shopify's bot filtering applies only to new data from October 7, 2025 onward and cannot classify older sessions. - Does Shopify catch every bot? No. Shopify says its classification is conservative, because it would rather miss some bots than label real customers as bots. Some bot sessions can still appear as human. Sources: - Shopify Help Center: Bot filtering in Shopify analytics and reports: https://help.shopify.com/en/manual/intro-to-shopify/bots/bot-filtering - Shopify Changelog (21 Sep 2026): Shopify Analytics session measurement improvements: https://changelog.shopify.com/posts/shopify-analytics-session-measurement-improvements - Shopify Help Center: Analytics updates: https://help.shopify.com/en/manual/reports-and-analytics/shopify-reports/analytics-updates - Shopify Changelog (27 Oct 2025): Filter out bot traffic in your sessions related reports: https://changelog.shopify.com/posts/filter-out-bot-traffic-in-your-sessions-related-reports ### Fix #19: Which products will sell out first on Black Friday? https://www.axioprax.com/fixes/which-products-sell-out-first-shopify (Inventory & reorders, published 2026-11-15) Short answer: Open Analytics > Reports, filter Category by Inventory and open Inventory remaining per product. Its Days of inventory remaining column estimates how long each variant lasts at your last 28 days of sales. Any best seller with fewer days left than your supplier's lead time will sell out first. Reorder those now with a purchase order. ### Why is a low-stock alert not enough before Black Friday? A low-stock alert tells you when a product has already dropped below a number you picked. That works on a normal week. Before a big sale it can come too late: if your supplier needs two weeks and the alert fires with five days of stock left, you will sell out before the new delivery lands. The better question is time, not units. How many days will this product last, and how many days does my supplier need? Shopify's built-in inventory reports give you the first number. You already know the second. ### How do I know which products are at risk? - A product sells steadily every day, but you reorder it only when someone notices the shelf is thin. - Your supplier is overseas or makes to order, so restocks take weeks. - Last year a best seller sold out during the sale and you lost the rest of the weekend. If any of these sound familiar, run the check below. You only need the products that bring in most of your revenue, so it is a short check. Steps: 1. Open the inventory reports: From your Shopify admin, go to Analytics > Reports, click the Category filter and click Inventory. On the mobile app, tap the menu icon, then Analytics > Reports, and filter Category by Inventory. 2. Open Inventory remaining per product: Read the Days of inventory remaining column. Shopify works it out as ending quantity divided by the variant's average daily sales over the last 28 days. 3. Add each supplier's lead time: Note how many days each supplier takes from order to stock on your shelf. We could not find a lead time field in Shopify's supplier details, so keep it in a simple sheet next to the report. 4. Flag anything that runs out before restock: Mark every variant whose days of inventory remaining is lower than its supplier's lead time. Those will sell out before new stock can arrive, even at normal pace. 5. Rank the flagged list by revenue: Open ABC product analysis in the same Inventory category and start with A-grade variants, the ones that make up about 80% of your revenue. Use Products by sell-through rate to see which of them have already sold the largest share of their stock. 6. Raise purchase orders: Go to Products > Purchase orders, click Create purchase order, select the supplier, add the flagged products and quantities, then click Save as draft. To send it, click More actions > Export PDF and email the PDF to your supplier, then click Mark as ordered once they confirm. 7. Check again before the sale: Open Inventory remaining per product again a few days before Black Friday on 27 November, because the 28 day window updates as you sell. ### How should I read the sell-through rate? Products by sell-through rate shows what share of your stock you sold: items sold divided by items sold plus items still in inventory. Shopify uses the most recent 30 day period available, which usually lags by about two days. A high rate on an A-grade product is a good reason to check its days of inventory remaining first. Shopify treats untracked products as having sold 0 units, so turn on tracking for anything you want to see here. ### When is this not the right fix? Be honest about the main limit. Both reports look backwards. Days of inventory remaining uses your last 28 days of sales, so it assumes Black Friday will sell at the same pace as an ordinary week. It will not. If your sale brings a spike, stock runs out faster than the report says, so add your own buffer on top of the lead time for anything you are promoting. It also cannot help a product with no recent sales, which shows N/A. And since 1 October 2026 the reports count On hand stock, which includes units already committed to orders, so the real days left can be lower than shown. If your counts are wrong to begin with, fix tracking before trusting any of this. ### What does it cost? Nothing extra on top of your plan. The inventory reports are included (Shopify lists them even on the Starter plan), and purchase orders are built into the Shopify admin under Products > Purchase orders. The cost is the time to note each supplier's lead time once. After that, the check is a short read of one report. ### Want this set up for you? Our Stocky Replacement kit (/store-kits/stocky-rescue) includes a supplier sheet and a reorder sheet that work with Shopify's own purchase orders, so lead times live in one place. To see how we rebuilt a weekly reorder routine on our demo store, read the demo build (/insights/stocky-rescue-demo-build). Questions: - How is this different from a low-stock alert? A typical low-stock alert fires when stock drops below a fixed number you set. This check compares how many days of stock you have left with how long your supplier takes, so it catches fast sellers while they still look well stocked. - Why does a product show N/A for days remaining? Shopify sets it to N/A when the variant had no sales in the last 28 days, because there is not enough data to predict. For a new product launching on Black Friday, estimate by hand. - Why did my stock numbers go up in these reports? Since 1 October 2026, Shopify's inventory reports measure On hand quantity instead of Available. On hand includes Committed and Unavailable stock, so figures can be higher than what you can actually sell. - Which plans have these reports? Shopify lists inventory reports on the Starter plan, and says the Basic plan comes with access to all reports. Sources: - Shopify Help Center: Inventory reports: https://help.shopify.com/en/manual/reports-and-analytics/shopify-reports/report-types/default-reports/inventory-reports - Shopify Help Center: Creating purchase orders: https://help.shopify.com/en/manual/products/inventory/purchase-orders/creating-purchase-orders - Shopify Help Center: Managing suppliers: https://help.shopify.com/en/manual/products/inventory/purchase-orders/managing-suppliers - Shopify Help Center: Basic plan: https://help.shopify.com/en/manual/intro-to-shopify/pricing-plans/plans-features/basic-shopify-plan - Shopify Help Center: Starter plan: https://help.shopify.com/en/manual/intro-to-shopify/pricing-plans/plans-features/shopify-starter-plan - Shopify blog: Retail holidays calendar 2026: https://www.shopify.com/blog/retail-holiday-calendar ### Fix #20: How do I set up instant answers in Shopify Inbox? https://www.axioprax.com/fixes/shopify-inbox-instant-answers (Support, published 2026-11-16) Short answer: Go to Sales channels > Inbox, click Chat settings, and in the Instant answers section click Create instant answer. Add a question and answer for your holiday shipping cutoff, returns and sizing, then click Save. Keep the default Track my order answer on, and set your Availability hours and an Outside of available hours first reply. ### What are instant answers, and why set them up before Black Friday? When a shopper opens chat in Shopify Inbox, they see a greeting and a short list of questions. If you are not using the Inbox agent, those questions are your instant answers. The shopper taps one and gets the answer you wrote, at once, even when nobody is at the desk. Anything they ask after that comes to your staff. In Black Friday week, many chats ask the same few questions: will it arrive in time, can I return it, which size do I need, where is my order. Writing those answers once, before the rush, means your team spends the week on the chats that need a person. ### How do I know I need this? - Last year you typed the same shipping cutoff reply again and again. - Chats arrive at night and wait until morning with no reply at all. - Your chat still shows only the default Track my order option. - Your holiday return window or delivery dates live on a page shoppers never find. If two of these sound familiar, give it an afternoon this week. Steps: 1. Turn on chat in your theme: Install Shopify Inbox, then go to Online Store, click Edit theme, select App embeds, then Online store chat. Turn on Online store chat and click Save. 2. Write your shipping cutoff answer: Go to Sales channels > Inbox, click Chat settings, and in the Instant answers section click Create instant answer. Put the shopper's question in the Question text box, your real last order date for holiday delivery in the Answer text box, and click Save. 3. Add returns and sizing answers: Repeat Create instant answer for your holiday return window and for sizing. Shopify also suggests answers from your store policies, so click a suggested answer, edit it to match your rules and click Save. 4. Put the busiest questions first: In the Instant answers section, drag the handle next to each answer so shipping and returns sit at the top, then click Save. Turn off any answer that does not apply this season instead of deleting it. 5. Check Track my order works: Track my order shows in chat whether or not your Inbox agent is active. Open the chat on your store and check it is there. Customers must sign in to review an existing order, so check that signing in works on your store before the sale. 6. Set your chat hours: Go to Sales channels > Inbox > Chat settings > Availability and use the Hours section to set a start and end time for each day. Use Apply these hours to all days if they are the same, then click Save. 7. Write an out of hours reply: In the Automated first reply section, click Turn on. Click Outside of available hours and, in the Message text box, say when you will reply and where to find shipping and returns answers. Click Save. ### How is this different from fixing where is my order emails? Fix #5 stops tracking questions before they are asked, with shipping emails and the order status page. This fix is about the chat window during your busiest week: the pre-sale questions about delivery dates, returns and sizes, plus clear hours so nobody waits in silence. ### When is this not the right fix? Instant answers are fixed text. They do not change themselves when your shipping cutoff passes, and we found no way to schedule them on Shopify's page, so put a reminder in your calendar to edit or turn off the cutoff answer the day it expires. A stale date is worse than no answer. If you have turned on the Inbox agent, the list of instant answers is replaced by its conversation starters, so your time is better spent on your store policies, which the agent answers from. And if most of your questions come by email rather than chat, start with Fix #5. ### What does it cost? Shopify Inbox is listed as free on the Shopify App Store. Allow an afternoon for setup. Most of that time goes on agreeing the real cutoff dates and return rules with whoever ships your orders. Questions: - How many instant answers can I add? Shopify says you can create as many as you want and display up to 100 to customers. For Black Friday, a short list of the questions you really get works better than a long one. - Why does Track my order ask customers to sign in? Shopify's page says that to review an existing order, the customer must sign in. Order lookup needs sign-in even if you do not require sign-in for staff chat. - I turned on the Inbox agent. Do instant answers still show? Not as the list of suggestions. Shopify says that when the Inbox agent is active it provides conversation starters and handles the responses, while Track my order stays available either way. The agent answers from sources such as your store policies and product catalog, so keep those up to date too. - Why did my out of hours reply not send? If the Inbox agent is set to Only when your staff is unavailable, Shopify sends an AI response outside your hours instead of the Outside of available hours reply. Sources: - Shopify Help Center: Set up instant answers for Shopify Inbox: https://help.shopify.com/en/manual/inbox/chat-settings-and-appearance/instant-answers - Shopify Help Center: Availability and automated first replies: https://help.shopify.com/en/manual/inbox/chat-settings-and-appearance/availability-and-first-reply - Shopify Help Center: Shopify Inbox settings (customer sign-in and online store chat): https://help.shopify.com/en/manual/inbox/chat-settings-and-appearance/settings - Shopify Help Center: Chat settings and appearance: https://help.shopify.com/en/manual/inbox/chat-settings-and-appearance - Shopify Help Center: Inbox agent persona: https://help.shopify.com/en/manual/inbox/agent-persona - Shopify App Store: Shopify Inbox: https://apps.shopify.com/inbox ### Fix #21: How do I control what AI shopping agents say about my store? https://www.axioprax.com/fixes/control-what-ai-agents-say-about-shopify-store (Support, published 2026-11-17) Short answer: Go to Sales channels > Agentic and click Knowledge Base under Sources. Approve or reject the facts Shopify scanned from your store, correct your return window and shipping details, add your holiday cutoff dates as an additional fact, then publish and save. Shopify calls these facts the source of truth for AI agents on non-product information. ### What is Shopify's knowledge base, and why should I care? Shoppers can ask an AI agent questions like "Can I return this after Christmas?" or "Do they ship to Ireland?" before they ever open your site. Shopify keeps a knowledge base of facts about your store for exactly these questions, and says: "These facts are the source of truth for AI agents for all non-product information about your store." Shopify builds most of it for you. It scans your public storefront and pulls in admin data such as your policies and delivery settings. The catch is the scan. Facts Shopify generates from your storefront "are unpublished until you publish them", and a scan can pick up an old FAQ page or a banner from last year. If nobody reviews them, agents may be working from gaps or stale details during your busiest weeks. This used to be a separate app called Shopify Knowledge Base. It now lives in the admin under Sales channels > Agentic. ### How do I know my store facts need attention? - You have never opened the Knowledge base page, so scanned facts may still be unpublished. - You changed your return window or shipping countries this year. - Customers ask about holiday delivery dates and nothing on your store states them in one place. - You used the old Knowledge Base app and have not checked what moved across. If any of these sound like your store, set aside time for a full review before Black Friday. Steps: 1. Open your knowledge base: From your Shopify admin, go to Sales channels > Agentic and, in the Sources section, click Knowledge Base. You need to be the store owner, or a staff member with the App and sales channel > Agentic permission or the App development > View apps developed by staff and collaborators permission. 2. Work through the suggested updates: For each suggested update from a store scan, click Approve & publish if it is right. If it is wrong, click the thumbs-down icon, choose a reason, click Submit, then click Save. 3. Correct your return window: Open the Returns & warranty category and read the return window and refund details. If the source icon points to your admin, such as Policies, fix it there; otherwise change the source to Manual, edit the field and click Save. 4. Check where you ship: Open the Fulfillment category, which covers the types of shipping and delivery your store offers. Make sure the countries you ship to, and any you do not, are stated clearly. 5. Add your holiday cutoff dates: In the Additional facts section, click Add fact, give it a clear title such as Holiday order cutoff dates and choose Paragraph. Enter your last order dates per shipping method, hover over Unpublished, click Publish, then click Save. 6. Publish what you have checked: Scanned facts stay unpublished until you publish them. Click Publish all on each category you have checked, or use Status > Published facts for everything, then click Save. ### How much control do I really have? You control the facts, not the final sentence. Shopify says knowledge base content "is never changed without your approval", and you choose what is published. But the agent still writes its own answer, so you cannot script its exact words. Keep facts short and plain: one return window, one list of countries, one set of dates. That leaves the least room for a muddled answer. After the holidays, unpublish the cutoff fact or update it, so agents are not quoting December dates in February. ### When is this not the right fix? If your store is password-protected, Shopify says the knowledge base cannot access your online store, so it only shows admin data. Custom storefronts and stores on the Agentic plan can add only additional facts. And if the problem is a product's size, price or description, the knowledge base will not help, because it covers non-product information. Fix the product listing instead. There is also a ceiling: the knowledge base holds a total of 1,000 facts, counting both the facts Shopify generates and the additional facts you add. ### What does it cost? Shopify's knowledge base help page does not list a charge, and the Knowledge Base app it replaces was listed as free in the Shopify App Store. The real cost is time: a first review takes the longest, then a quick check whenever your policies or delivery change. Questions: - Which AI agents use my knowledge base? Shopify says the published knowledge base is publicly available through the Storefront MCP and can be used by AI channels including Shop and third-party platforms. The help page does not list every agent by name. - I used the Knowledge Base app. Do I lose my FAQs? If the app had 1,000 or fewer facts, Shopify transfers your overrides and added facts automatically and publishes them in the new knowledge base. If it had more, you need to add the extra ones by hand before the app is deprecated on January 15, 2027. - I updated my shipping page. Why has the fact not changed? Shopify scans your store regularly and offers new information as a suggested update. To pick up a change sooner, request a rescan from the Knowledge base page, which Shopify says can take up to 48 hours, or edit the fact directly. - Does this change what agents say about my products? No. The knowledge base covers non-product information like policies, contact details and delivery. Product details come from your catalog, so fix those on the product itself. Sources: - Shopify Help Center: Knowledge base (agentic storefronts): https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/knowledge-base - Shopify Help Center: Managing agentic storefronts: https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/agentic-home - Shopify Help Center: Managing FAQs in the Knowledge Base app: https://help.shopify.com/en/manual/promoting-marketing/knowledge-base/managing-faqs - Shopify News: Introducing Shopify Agentic Storefronts: https://www.shopify.com/news/winter-26-edition-agentic-storefronts - Shopify App Store: Shopify Knowledge Base: https://apps.shopify.com/shopify-knowledge-base ### Fix #22: What is a Shopify Payments reserve, and when is the money released? https://www.axioprax.com/fixes/payments-reserve-release-shopify (Payouts & holds, published 2026-11-18) Short answer: A reserve is money Shopify Payments holds back for a set period to cover chargebacks and refunds. It is either a fixed amount or a percentage of each transaction. You see the amount under Reserve in Finance > Payouts. The terms and expiry date come by email, and funds not needed are included in upcoming payouts after expiry. ### What is a Shopify Payments reserve, and why should I care? Shopify describes a reserve as "a temporary hold on a portion, in some cases a full amount, of transactions processed through Shopify Payments for a specified period." It is there to cover chargebacks and refunds. If a customer disputes a payment, the amount and any dispute fee come out of your account, and the reserve makes sure that money is there. Shopify uses two types. A fixed amount reserve sets aside a specific sum for a set time. A percentage-based reserve holds back a share of every transaction for a set time and pays the rest out as normal. Shopify's own example is a 10% reserve for 120 days: 10% of each sale is held and 90% goes out in your regular payouts. Shopify says a reserve can follow a risk assessment that looks at things like extended billing cycles, more chargebacks, a higher refund rate, long delivery times such as pre-orders, or a sudden jump in sales volume. That last one matters in November and December. ### How do I know I have a reserve? - A Reserve section appears on your Payouts page. Shopify only shows it when your account has reserved funds. - You received an email from Shopify setting out reserve terms. - Your payouts are smaller than your sales, and the breakdown shows a Reserved funds amount. Your total balance is your Pending amount plus your Reserve amount. Only Pending is heading to your bank in upcoming payouts. Steps: 1. Open your Payouts page: From your Shopify admin, go to Finance > Payouts. Pending is money due in upcoming payouts, and Reserve, shown only when your account has reserved funds, is money you cannot access until the reserve period ends. 2. Find the reserve terms email: Shopify communicates the terms of a reserve by email, so search the inbox you use for Shopify and the spam folder. Shopify sends account hold emails to the store owner address. To check it, go to Settings > Users and permissions and click the name under Store owner. 3. Note the type, period and expiry date: From the email, write down whether it is a fixed amount or a percentage-based reserve, how long it runs, its expiry date and any deadline for appealing. Put the expiry date in your calendar. 4. Review your reserve transactions: In Finance > Payouts, click Reserve. Reserve funds are negative transactions taken from your payouts, and Released funds are positive transactions that become available for payout after the reserve is complete. 5. Export the reserve for your books: On the same page, click the export icon, then click Export reserve transactions to save a CSV. After everything is released, click View transactions on the Payouts page and filter by the Reserved Funds transaction type instead. 6. Check each payout breakdown: In Finance > Payouts, click the payout date, then the payout. Reserve transactions are included in the Reserved funds amount of the breakdown, so you can see how much each payout lost to, or gained from, the reserve. 7. Track it over a month: Go to Finance > Documents and click Shopify Payments activity report. The Reserves & Holds group lists reserved funds and the reversals that release them back to your balance, with a 3-day data delay. ### When is the money released? Each reserve has an expiry date, and Shopify sends the reserve terms by email. Before it expires, Shopify reviews your account and decides whether to lift, maintain, reduce or increase the reserve. If no extension is needed, the reserve amount is included in upcoming payouts, and Shopify says it "might take several business days" to be processed. At the end of the period, reserve funds no longer needed for refunds or disputes are returned in full. Shopify does not promise an earlier date, so plan your cash flow around the expiry date, not before it. If you think the reserve is wrong, reply to the reserve email within the timeframe it gives. Some reserves cannot be appealed. If the email lists requirements instead, Shopify removes the reserve automatically once you meet them. ### When is this not the right fix? If your whole payout has stopped and the notice talks about information, bank details or product eligibility, that is an account hold, not a reserve. Read our payout hold fix instead. A creditor hold is different again: Shopify says it cannot remove that type, only the payment processor can, after a release from your creditor. And if there is no Reserve section, a smaller payout is more likely down to refunds, chargebacks, fees, adjustments or a negative balance. ### What does it cost? Shopify's reserves page does not mention a fee for having a reserve. The cost is cash flow: money you have earned but cannot spend on stock or ads until the reserve ends. The time cost is small: a regular look at the Payouts page and a fresh export when the reserve changes. Questions: - Can I still take payments while I have a reserve? Usually, yes. Shopify says reserves generally don't stop you from accepting payments through Shopify Payments, unless it tells you otherwise. - What is the difference between a fixed and a percentage-based reserve? A fixed amount reserve sets aside a specific sum for a defined duration. A percentage-based reserve withholds a percentage of each transaction for a specified period, and the rest is paid out as normal. - Can the reserve be extended? It can. Before a reserve expires, Shopify reviews your account and decides whether to lift, maintain, reduce or increase it. Shopify says the review is based on risks such as chargebacks or refunds. - Why do I see a negative reserve transaction after an early release? If a reserve is released early, you might receive one bulk positive reserve transaction. When the original expiry date arrives, the original positive transactions are paid out with equal negative ones to account for the early release. Sources: - Shopify Help Center: Overview of reserves in Shopify Payments: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/reserves - Shopify Help Center: Viewing and exporting your Shopify Payments payout details: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/view-details - Shopify Help Center: Shopify Payments activity report: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/payouts-activity-report - Shopify Help Center: Shopify Payments account holds: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/account-holds - Shopify Help Center: Lower or missing payouts: https://help.shopify.com/en/manual/payments/shopify-payments/payouts/lower-or-missing-payouts ### Fix #23: How do I add a delivery date range to Shopify order confirmation emails? https://www.axioprax.com/fixes/add-delivery-dates-order-confirmation-shopify (Support, published 2026-11-19) Short answer: Turn on estimated delivery dates in Settings > Shipping and delivery > Estimated delivery dates, choose your fulfillment time and add transit time to your flat rates. Shopify then shows the estimate in the Order confirmation email. If you edited that template before, add Shopify's consolidated_estimated_delivery_time snippet back, then add a return policy link. ### What is a delivery date range, and why put it in the confirmation email? A delivery date range tells the customer when the parcel should arrive, for example Thursday to Friday, rather than a vague "3 to 5 days". Shopify calculates it for you once you set two things: how long you take to fulfill an order, and how long the carrier takes to deliver it. Shopify adds your fulfillment time to the carrier's transit time and shows the result as a date range. The Order confirmation is the email Shopify sends after a customer places an order. It is the one they search for when they start to wonder where the parcel is. In Fix #5 we covered shipping emails and the order status page, which help once an order has shipped. This fix covers the days before that, when there is no tracking yet and the confirmation email is all the customer has. ### How do I know my confirmation email is missing it? - Place a test order and read the confirmation. If it shows the shipping method name but no "Estimated delivery" line, it is missing. - You or an app edited the Order confirmation template in the past. Shopify warns that delivery dates might not be included in a customized template by default. - Checkout shows a transit time like "3 to 4 business days" but never actual dates. - Most "Where is my order?" emails arrive before the order has shipped. Steps: 1. Turn on estimated delivery dates: From your Shopify admin, go to Settings > Shipping and delivery > Estimated delivery dates, click Estimated delivery dates and select Manual (or Automated if your store is eligible). Choose your Fulfillment time from the drop-down menu and click Save. 2. Add transit time to your flat rates: Go to Settings > Shipping and delivery, open a shipping profile, click the menu next to a shipping option and click Edit shipping option. Pick a Transit time, click Done, then Save; carrier-calculated rates include transit time already. 3. Open the Order confirmation template: Go to Settings > Notifications and, in the Orders section, click Order confirmation. Click Edit code to see the Email subject and Email body (HTML). 4. Put the delivery estimate back if it is missing: If you customized this template, search the body for consolidated_estimated_delivery_time. If it is not there, paste the snippet from Shopify's manual delivery dates guide after the {% if requires_shipping %} line. For Automated dates, Shopify gives a separate snippet that replaces the shipping_method.title line. 5. Type a range for rates without dates: For shipping rates where Shopify shows no date, such as rates from a third-party app, add your own sentence per rate using a condition on shipping_method.title. Use the rate names exactly as they appear in your shipping settings. 6. Add a link to your return policy: Under the delivery estimate, add a link to {{ shop.url }}/policies/refund-policy, the direct address Shopify gives for your return policy. Check the policy itself first in Settings > Policies. 7. Preview, test and save: Click Preview to check your unsaved changes, then Send test email, which goes to the account you are logged in with. If it looks right, close the preview and click Save. ### What if Shopify cannot show a date for my rates? Then the range is text you type yourself. Shopify's estimate comes from your delivery date settings, so for those rates you write a sentence for each shipping rate name. Here is a simple shape, with your own rate names and times: {% if shipping_method.title == 'Standard Shipping' %} Usually delivered 3 to 5 business days after we dispatch. {% elsif shipping_method.title == 'Express' %} Usually delivered 1 to 2 business days after we dispatch. {% endif %} Our return policy One honest limitation: Shopify translates the rate name in customer-facing notifications when the customer checked out in another language, so a condition written in English may not match. If you sell in several languages, test an order in each one. ### When is this not the right fix? If your real problem is slow dispatch, a date in an email will only make the delay more visible. Fix the fulfillment time first. Shopify's own policy guidance says not to promise delivery or shipping times you cannot provide, and that applies to typed ranges too. Also, delivery dates depend on your origin country and rate type, so check Shopify's availability table before you plan around them. ### What does it cost? Shopify lists manual delivery dates as available on all plans. Automated delivery dates need the Basic plan or higher and at least one fulfilled domestic order. Editing notification templates is part of your Shopify admin, though you first need to confirm your sender email address. The time cost is a careful test order for each shipping rate you offer. Questions: - Does Shopify add delivery dates to the confirmation email on its own? Yes, once delivery dates are set up and the order is eligible. Shopify says the estimate shows at checkout, in order confirmation emails and on the Thank You and Order Details pages. A template you customized earlier may not include it until you add the snippet. - Why does my checkout show no delivery dates at all? Shopify lists several reasons: dates do not work with shipping rates from third-party apps or with negotiated FedEx-calculated rates, and if checkout shows a mix of eligible and ineligible rates, no dates are shown. Availability also depends on your origin country and rate type. - Do weekends and holidays count in the estimate? For manual delivery dates, Shopify counts business days as Monday to Friday and uses a 12 pm cutoff in the local time zone of the shipping origin. Its worked example shows weekends and a holiday not counted as fulfillment time. - Can I undo my template changes? Yes. Open the template, click Revert to and choose Previous or Default, then click Save. Reverting changes only the template content, not your logo or colors. Sources: - Shopify Help Center: Setting up manual delivery dates: https://help.shopify.com/en/manual/fulfillment/setup/delivery-expectations/manual-delivery-dates - Shopify Help Center: Setting up automated delivery dates: https://help.shopify.com/en/manual/fulfillment/setup/delivery-expectations/automated-delivery-dates - Shopify Help Center: Getting started with fulfillment time and delivery dates: https://help.shopify.com/en/manual/fulfillment/setup/delivery-expectations/overview - Shopify Help Center: Setting up transit time for your flat shipping rates: https://help.shopify.com/en/manual/fulfillment/setup/shipping-rates/transit-time - Shopify Help Center: Customizing email notification templates: https://help.shopify.com/en/manual/fulfillment/setup/notifications/customizing-notification-template - Shopify Help Center: Notifications variables reference: https://help.shopify.com/en/manual/fulfillment/setup/notifications/email-variables - Shopify Help Center: Adding store policies: https://help.shopify.com/en/manual/checkout-settings/refund-privacy-tos ### Fix #24: How do I flag new orders from a customer who already filed a chargeback? https://www.axioprax.com/fixes/flag-orders-after-chargeback-shopify (Fraud & bots, published 2026-11-20) Short answer: Shopify Flow has a Dispute created trigger for orders that use Shopify Payments. Pair it with the Add customer tags action to tag the customer, such as dispute-review, when a chargeback or inquiry opens. Then install the template "Hold order and add order tag when a customer has a certain tag" so their next orders wait for review. ### Why does a chargeback not stop the same customer ordering again? A chargeback is about one order. Whether you win it, lose it or accept it, nothing in that process changes how your store treats the customer next time. Their account still works, their email still goes through checkout, and their next order lands in your Unfulfilled list like any other. Sometimes that is fine. A dispute can be a genuine mix-up, such as a parcel the cardholder thinks never arrived. But if you ship the next order without a second look, you can end up with two disputes from the same person instead of one. Shopify Flow has a trigger for exactly this. Shopify says the Dispute created trigger "starts a workflow when a chargeback or inquiry dispute is created for an order in your store that uses Shopify Payments." It works with the Add customer tags action, so the customer is marked the moment a dispute opens. A second workflow then checks for that tag on every new order and holds it. ### How do I know I need this? - You have had a chargeback or inquiry and could not say whether that customer has ordered since. - You only find a customer's dispute history by scrolling back through their past orders. - Your packing team ships everything in Unfulfilled without checking who the customer is. - You take payments through Shopify Payments, which this trigger needs. If any of these sound familiar, the setup below takes two workflows and one tag. Steps: 1. Create the dispute workflow: From your Shopify admin, go to Apps > Flow and click Create workflow. Click Select a trigger and choose Dispute created. 2. Tag the customer: Click + to add a step, choose Action, then Add customer tags, and enter a neutral tag such as dispute-review. Name the workflow and click Turn on workflow. 3. Tag customers from earlier disputes: The trigger only runs for new disputes. Go to Orders, click Search and filter, add the Chargeback and inquiry status filter and select Open, then open each order's customer, add the same tag in the Tags section and click Save. 4. Install the hold template: In Apps > Flow, click Create workflow, then Browse templates, and select Hold order and add order tag when a customer has a certain tag. Click Install. 5. Point it at your tag: Change the customer tag in the template's condition to dispute-review, and change its order tag to one your team will search for. Fill in Reason and Reason notes in the Hold fulfillment order action, then click Turn on workflow. 6. Test both workflows: With Shopify Payments in test mode, create a test dispute using the dispute card numbers from Shopify's payments test mode page. Check that the customer got the tag, then place a second test order as that customer and confirm it goes On hold. 7. Review held orders every day: Go to Orders, click Search and filter > Tag and select your order tag. Open each order and, if it looks genuine, click Release hold in the On hold section; if not, cancel it. ### What should we do with a held order? Look before you decide. Open the earlier disputed order and read the reason. An item not received dispute from a customer who later confirmed delivery is very different from a fraudulent one. If the new order looks genuine, release the hold and it goes back to Unfulfilled. If it does not, cancel it. Shopify says orders that are on hold "are excluded from the list of Unfulfilled orders", so held orders are easy to forget. Pick one person to check the order tag each day. When a dispute is settled and you are happy to trust the customer again, remove the tag from their customer page. Flow also has a Remove customer tags action if you would rather do it in a workflow. ### When is this not the right fix? The honest limit: the tag sits on one customer record. If the same person checks out with a different email, the new order belongs to a customer without the tag and will not be held. For that, block the email, address or IP with Fraud Control, as in our repeat fraudster fix. The trigger also only covers orders that use Shopify Payments, so disputes through another payment provider will not tag anyone. A hold stops shipping, not payment. And none of this changes how the dispute itself ends. Respond to that separately, as in our chargeback fix. ### What does it cost? Nothing extra. Shopify Flow is listed as free in the Shopify App Store, and the trigger, actions and template used here are part of it. The real cost is time: someone has to review held orders each day, and a genuine customer may wait a little longer for their parcel. Questions: - Does the tag go on even if I win the chargeback? Yes. The workflow runs when the dispute is created, not when it is decided, so the tag goes on either way. Removing it after a win is your call, on the customer page or with Flow's Remove customer tags action. - Can I do this without a customer tag? Yes. Shopify's Flow template "Hold fulfillment orders for customers with chargebacks" checks the same customer's orders from the past 60 days for a chargeback when an order is ready to fulfill. It only looks back 60 days, while a tag stays until you remove it. - Flow says it added the tag, but I cannot see it. Why? Shopify says another app or a staff member can overwrite a tag Flow just added, often because they react to the same events. Shopify suggests asking the app how to avoid it, or adding a Wait step before Flow adds the tag. - Will the customer see the tag? Tags are not displayed to customers, but Shopify warns that installed apps which can view customers could show customer tags to them. Keep the tag neutral, such as dispute-review, rather than something like fraudster. Sources: - Shopify Help Center: Dispute created (Flow trigger): https://help.shopify.com/en/manual/shopify-flow/reference/triggers/dispute-created - Shopify Help Center: Add customer tags (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/add-customer-tags - Shopify Help Center: Hold fulfillment order (Flow action): https://help.shopify.com/en/manual/shopify-flow/reference/actions/hold-fulfillment - Shopify Help Center: Creating workflows manually or from a template: https://help.shopify.com/en/manual/shopify-flow/create/create-workflow - Shopify Help Center: Managing chargebacks in the Shopify admin: https://help.shopify.com/en/manual/payments/chargebacks/chargebacks-in-admin - Shopify Help Center: Marking your order as on hold: https://help.shopify.com/en/manual/fulfillment/fulfilling-orders/holding-fulfillments - Shopify Help Center: Creating and using tags in Shopify: https://help.shopify.com/en/manual/shopify-admin/productivity-tools/using-tags ## Insights (27 live) ### Why AI Automations Fail in Production https://www.axioprax.com/insights/why-ai-automations-fail-before-they-ship (2026-08-24) AI automation works in the demo and fails in production. The four root causes are scope, data access, edge cases and adoption, and here is how to prevent each. The demo goes well. Data flows in, the AI makes sensible decisions, the right emails go out. Six weeks after launch the workflow is switched off, and nobody can quite say why. We see this pattern more than any other. The technology usually works. What fails is everything around it: a process nobody fully wrote down, data the automation couldn't reach, cases nobody planned for, and people who were never shown what changed. Here are the four causes, how to spot each one early, and the checks to run before you build. ### Why the demo isn't the test A demo is built to succeed. The sample data is clean, the steps run in order, and the awkward cases have quietly been left out. Real operations aren't like that. Take a typical lead workflow: a form submission is enriched with company data, written to the CRM and followed by a personalised email. The demo never shows the forms where the company field is blank, the leads from personal email addresses that can't be enriched, the CRM field someone renamed last week, or the email that went out twice because a webhook fired twice. Production shows all of them in the first week. ### 1. The process was never written down The most common cause has nothing to do with AI. The automation was built for a process nobody fully understood, so it was built on assumptions, and assumptions break. Before anyone configures a single step, write the process down as it actually runs today, not as it's supposed to run. Who does what, in which tool, and what they do when something looks wrong. The workarounds matter most, because they're where the real rules live. If the team can't agree on the steps, you've found the first job, and it isn't automation. ### 2. Data access turned up late Automations live on access. An enrichment step needs an account with a data provider. Invoice processing needs readable documents, not scans of scans. A triage agent needs permission to read the support inbox. In a demo all of this is already set up. In a real business, API keys have limits, IT has approval steps, documents arrive in a dozen formats, and a permission request can sit for weeks. List every system the automation will read or write before building starts, test the access with real records, and start any approval on day one. ### 3. Nobody planned for the exceptions A process that looks like five steps usually has dozens once you count the exceptions. An invoice workflow handles standard invoices well. Then a two-page invoice arrives, or a credit note, or one in another currency, or a photo of a crumpled receipt. An automation without a plan for these doesn't stop politely. It skips the record, or worse, processes it wrongly, and nobody notices until a supplier chases a payment. Go through the exceptions the team handles today and give each one an explicit answer. ### 4. Nobody showed the people Automation changes what people do. The person who used to type invoices into a spreadsheet now reviews the ones the automation couldn't read. The salesperson who used to research every lead now gets a filled-in record and needs to know where the data came from. When nobody explains the change, people fall back on the old way. The automation runs in parallel and doubles the work, or someone switches it off after it does something surprising and nobody switches it back on. Treat the handover as part of the build: who uses the output, what they need to trust it, and who owns the queue of things it couldn't do. ### Five checks before you build - The process is written down, workarounds included, in plain language. - Data access is confirmed with real records, not test accounts. - Every exception has an answer: handle it, send it to a person, or stop and alert. - Success is defined, and today's number is written down, whether that's hours spent, error rate or time to respond. - The exception owner is named, and they know it's their job. Check all five before you build. If one is missing, that gap is the most likely place for the automation to break once real data and real users reach it. ### After launch Run the automation alongside the manual process for the first two weeks and compare the results. Switch the manual process off only when they match. After that, set an alert for runs that fail and a weekly look at the exception queue, because an automation that stops quietly looks exactly like one that has nothing to do. If you're still choosing what to automate first, our guide to automation ROI (/insights/ecommerce-automation-roi) shows how to pick by payback. For systems where an AI makes its own decisions, building AI agents that work in production (/insights/designing-ai-agents-that-work-in-production) covers the extra guardrails they need. ### n8n vs Zapier (2026): Pricing, Limits and When to Switch https://www.axioprax.com/insights/n8n-vs-zapier-honest-comparison (2026-08-25) n8n counts workflow runs, Zapier counts every action that succeeds. Cost, security, self-hosting and upkeep compared honestly, with prices read on each vendor's page. Most n8n vs Zapier comparisons end with an affiliate link. This one doesn't. We have no deal with either company, and we think the honest answer is simpler than the usual feature tables make it look: the two tools bill for different things. Once you see what each one counts, the choice for your business usually makes itself. The prices here come from each vendor's own pricing page, read on 20 October 2026. They change, so check them again before you commit to anything. ### What each one actually counts Zapier counts tasks. A task is one action that worked: a row added, a contact updated, a message sent. The trigger that starts the Zap is free, so are Filter and Paths steps, and an action that fails doesn't count either. Zapier's help centre (https://help.zapier.com/hc/en-us/articles/8496196837261-Tasks-in-Zapier) spells all of this out. So a Zap with one trigger and four actions uses four tasks every time it runs. n8n counts executions. An execution is one run of the whole workflow, start to finish, and it makes no difference whether that run has three steps or thirty. The same four-action workflow on n8n costs one. That single difference explains most of what people say about the two. Zapier gets expensive when workflows get long. n8n doesn't, but it asks more of you in other ways, which we'll come to. ### The plans, side by side Zapier's free plan gives you 100 tasks a month and only two-step Zaps. Professional starts at $19.99 a month and Team at $69, and how many tasks you get depends on the tier you pick (Zapier pricing (https://zapier.com/pricing)). Worth knowing before you sign up: when you hit your limit, Zapier moves you onto pay-per-task billing at a higher rate, unless you've turned that off. If you have, your Zaps pause. n8n Cloud starts at €20 a month, billed annually, for 2,500 executions. Pro is €50 a month for 10,000 (n8n pricing (https://n8n.io/pricing/)). You can also run n8n on your own server with the free Community Edition, and that's where most of the savings people talk about come from. It isn't free in time, though. Someone has to keep it updated and backed up. Zapiern8n What it countsEach action that succeedsEach run of the whole workflow First paid planProfessional, from $19.99 a monthStarter, €20 a month billed annually, 2,500 runs Free option100 tasks a month, two-step ZapsCommunity Edition on your own server Who runs itZapiern8n Cloud, or you When you hit the limitPay-per-task at a higher rate, or Zaps pauseMove up a plan ### One workflow, priced both ways Take a flow almost every growing business with an online shop ends up wanting. When an order is paid, tag the customer in Shopify, add a line to the orders sheet, drop a note in the team's Slack channel, and create or update the contact in HubSpot. One trigger, four actions. Say you do 1,000 paid orders a month. Swap in your own number. On Zapier that's 4,000 tasks. On n8n it's 1,000 executions, which is 40% of the Starter plan. Now the bit that matters. Six months later someone asks for a WhatsApp shipping message on top. Zapier goes to 5,000 tasks. n8n stays at 1,000. Every step you add to a Zapier workflow lands on the bill, and on n8n it doesn't. We're not going to quote you a Zapier price for 4,000 tasks. The pricing page doesn't list one per tier, and the selector changes, so put your own volume in and look. Also, this is a model. Retries, replays and branches move the numbers, and on the Zapier side they almost always move them up. If you want the full arithmetic, including Make and what polling costs on each platform, we wrote it up in tasks, credits and executions (/insights/tasks-credits-and-executions). ### Where Zapier wins We'll be fair to Zapier, because it earns it. If nobody on your team wants to think about servers, it's the obvious pick. The editor is easier for someone building their first automation, the app library is huge, and short two- or three-step jobs cost very little. If all you want is each new order landing in a Google Sheet, Zapier is fine. More than fine. ### Where n8n wins n8n pays off when workflows get long or messy. Branches, loops, a bit of code to clean up a supplier's CSV before it goes anywhere near your stock levels. It can call any API through its HTTP Request node, so you're not waiting for someone to build an integration for the small 3PL you happen to use. The other reason is data. Self-hosted n8n keeps everything on a server you control, and for some businesses that settles it. Just go in clear-eyed: updates, backups and who has access become your job, or your developer's. ### Is n8n safer than Zapier? For sensitive customer data, self-hosted n8n is usually the safer choice, because the data stays on a server you control. Zapier is the safer choice only if nobody on your team can keep a server updated, backed up and locked down. Self-hosted n8n keeps your workflow data on a server you control, so customer records never pass through another company's platform. That is often the deciding point for teams with GDPR obligations or clients who ask where their data goes. The trade-off is responsibility. On your own server, updates, backups, access control and monitoring are yours to handle. With Zapier or n8n Cloud, the vendor runs the platform and your data goes through their servers on their terms. Neither is safer by default; one is safer if someone is looking after it. ### Moving from Zapier to n8n There is no one-click import. Each Zap is rebuilt as an n8n workflow, run side by side with the original, and switched over once the outputs match. Short Zaps take minutes. Long ones with branches and filters take longer, and they are usually the ones costing you the most. Move the expensive Zaps first and leave cheap two-step ones alone if they work. We do this as a fixed-price Zapier to n8n migration (/services/zapier-to-n8n-migration). ### So which one? Our short version. If your workflows are short and nobody wants to look after anything, use Zapier. If your workflows are long, your order volume is climbing, or your data needs to stay on your own server, use n8n. When we build, we end up on n8n when a workflow has a lot of steps or the data has to stay put. If you want someone to build and look after your n8n workflows, see our n8n automation service (/services/n8n-automation). ### If you run a Shopify store, look at Flow first If you run a Shopify store, check what you already have. Shopify Flow (https://help.shopify.com/en/manual/shopify-flow) is free on the Basic, Grow, Advanced and Plus plans, and a lot of store automation never needs to leave Shopify at all. Tag a customer after their third order. Hold an order that looks risky. Tell someone when a product runs low. Flow handles all of that. The catch is that Flow lives inside Shopify. Its Send HTTP request action, which is how it talks to other apps, only comes with Grow and above. So the moment a job needs your supplier sheet, your CRM or a carrier, that's when n8n or Zapier earn their place. We follow the same rule in our own kits. Fraud Hold, for example, is just Shopify Flow rules: hold the fulfilment order, tell a named person, wait for them. Nothing in that job needs a second tool, so it doesn't get one. Not sure where your store falls? The Store Ops Audit (/store-kits) goes through the automations and apps you already run and tells you what to fix first. ### How to Build Your First n8n Workflow https://www.axioprax.com/insights/build-your-first-n8n-workflow (2026-08-25) Build your first n8n automation from scratch: choose a trigger, connect apps, test with real data and switch it on. No prior automation experience needed. Your first n8n workflow is easier than it looks, as long as you build it in the right order. This guide walks through one small, useful example from start to finish: when someone fills in your website form, add them to a Google Sheet and post a message to your team. Along the way it covers the habits that keep workflows working after the first week. ### Before you start You need somewhere to run n8n. n8n Cloud is the quickest start, and n8n also publishes a free self-hosted version you can run on your own server. On 18 September 2026, n8n's Starter cloud plan was 20 EUR a month billed annually with 2,500 workflow executions included (n8n pricing (https://n8n.io/pricing)). Plans change, so check the page before you sign up. If self-hosting appeals, decide that separately, because you take on updates, backups and security yourself. Then gather access to the apps you'll connect. For this example that's your Google account for Sheets and your Slack workspace or email. Having these ready before you build saves breaking off halfway to hunt for logins. ### The example we'll build ### 1. Pick the trigger Every workflow starts with a trigger: the event that makes it run. There are three you'll use most. For a form, use a Webhook trigger if your form tool can send one, which most can. n8n gives you a test URL and a production URL. Point the form at the test URL while you build, and switch to the production URL when you go live. Forgetting that switch is the most common reason a first workflow "works in testing" and then never runs. ### 2. Build one node at a time Send one real test submission through the form so the webhook has data to show. Then add the next node, the Google Sheet, and run just that node. Check the row appeared. Then add Slack and run that. Testing each node as you add it turns debugging into a short question: which node did the last thing I expected? You can also pin the test data on the trigger, so you don't have to submit the form again every time you change something further down. Data passes from node to node as fields. When you fill in the Sheet node, use the expression editor to pick fields from the webhook, such as the name and email, rather than typing their names by hand. Its suggestions show exactly what's available, which avoids the most common beginner mistake: referring to a field that doesn't exist. ### 3. Handle the messy cases Real forms arrive with blank fields, spam and odd characters. Add an IF node after the trigger that checks the email isn't empty. Send good submissions down the main path, and send the rest to a separate tab in the sheet so a person can look at them. Nothing is lost, and the team channel doesn't fill with junk. ### 4. Test with three different inputs Run the whole workflow with at least three test submissions: a normal one, one with a blank email, and one with a long message full of punctuation. Look at each node's output for empty fields, error messages and anything that ran when it shouldn't have. Problems usually show up on the second or third test, not the first. ### 5. Go live, and set a safety net Switch the form to the production URL and turn the workflow on. Then set up an error workflow: a second, small workflow that starts with n8n's Error Trigger node and sends you an email or Slack message whenever another workflow fails. Without it, a broken workflow stops quietly and you find out from a customer. Check the list of executions after the first few real submissions to confirm they ran as expected. After that, the error workflow does the watching. ### What to build next Once the first one works, the second goes much faster, because the patterns repeat: a trigger, a check, an action, a safety net. For store ideas, see five n8n e-commerce automations (/insights/n8n-ecommerce-automation-workflows). For what this example looks like as a full enquiry process, with an owner and a follow-up, see what happens between the form and the first follow up (/insights/between-the-form-and-the-first-follow-up). And if you're still choosing a platform, n8n vs Zapier (/insights/n8n-vs-zapier-honest-comparison) compares them honestly. ### 5 n8n E-commerce Automations Every Store Needs https://www.axioprax.com/insights/n8n-ecommerce-automation-workflows (2026-08-26) Abandoned cart recovery, low stock alerts, post-purchase emails, review requests and weekly sales reports: five n8n workflows for e-commerce stores. These are the five store automations people most often want built in n8n. For a couple of them the honest answer is that Shopify already does the job, and n8n only earns its place once you need something Shopify can't do. Here's each one, what you get for free, and what the n8n version adds. The rule we use is simple. If a job starts and ends inside Shopify, build it in Shopify Flow first. Move it to n8n when it has to read or write another system, needs more data than Flow can fetch in one go, or needs logic that would turn a Flow workflow into a maze. ### 1. Abandoned cart recovery For most stores, don't build this one in n8n. Shopify Messaging and Flow both have abandoned checkout automations, and email apps like Klaviyo have their own. They already know who agreed to marketing and stop when the customer orders. An n8n version makes sense in one situation: you want to reach customers on a channel your email tool doesn't cover, or you need to check something in another system before sending. Shopify doesn't send a webhook when a checkout is abandoned, so n8n reads abandoned checkouts from the Admin API on a schedule. Before every message it should look up whether that customer has ordered since and whether they agreed to be contacted. Skip either check and you'll message people who already paid. ### 2. Low-stock alerts that count days, not units Flow's low-stock template alerts you when a product drops below a number you choose. That works for a short catalogue. It struggles when ten units means a month for one product and a day for another, and Flow's get data actions only handle lists of up to 100 items at a time (Shopify Help (https://help.shopify.com/en/manual/shopify-flow/advanced-workflows)). The n8n version runs each morning. It pulls stock on hand and the last 28 days of sales for every variant, works out how many days of stock are left, and compares that with how long your supplier takes to deliver. One trap: if a product was out of stock for part of those 28 days, its average looks lower than real demand, and the alert comes too late. Count only the days it was actually on sale, or at least flag products that hit zero recently. For a worked example of the reorder side, see our Stocky replacement demo build (/insights/stocky-rescue-demo-build). ### 3. Post-purchase updates when a parcel stalls Shopify already sends the order confirmation, the shipping confirmation and, where the carrier reports it, "out for delivery" and "delivered" emails. Don't rebuild those. What n8n adds is noticing when nothing happens. It checks fulfilled orders daily and flags the ones with no carrier scan a few days after the label was made, or no movement for longer than the route usually takes. The customer gets a short note before they have to ask, and your team gets a list to chase. Because carriers report differently, test this on your real carriers before trusting it, and keep the messages plain: what you know, what you're doing, when you'll update them. ### 4. Review requests after delivery Your reviews app can usually send the request itself, timed from fulfilment or delivery. Use n8n when you want rules the app doesn't offer: wait a few days after a delivered event, skip orders with an open return or refund, skip orders that arrived late, and send each customer one request at most. The sent log matters more than it looks. Webhooks can arrive twice and workflows get rerun after a fix. Writing each order ID to a sheet or table, and checking it before sending, is what stops a customer getting the same request three times. ### 5. A weekly sales report people actually read Flow can now send a weekly email with numbers from your analytics, using a scheduled trigger and the Get analytics data action. It sends values rather than a formatted report, which is fine for five or six figures (Shopify Help (https://help.shopify.com/en/manual/shopify-flow/reference/actions/get-analytics-data)). The n8n version is worth it when you want context: this week against last week and the same week last year, ad spend next to sales, refunds by reason, and one line on anything unusual. It can write the numbers to a Google Sheet as well, so you build up a history without exporting anything by hand. ### Make it reliable before you make it clever Most broken store automations don't fail loudly. They stop, and nobody notices for weeks. Three habits prevent that: - An error workflow. n8n's Error Trigger node starts a separate workflow when another one fails. Point it at an email or Slack message so a failure reaches a person. - A log of what was sent. Record order IDs before acting on them, so a rerun or a duplicate webhook can't send twice. - Room for Shopify's limits. Shopify limits how fast apps can call its API. Fetch in pages and slow down when it tells you to, rather than assuming every call succeeds. ### What it costs to run n8n counts workflow runs, not steps, which suits these jobs because most of them are a handful of runs a day. Our comparison of tasks, credits and executions (/insights/tasks-credits-and-executions) works through what the same workflow costs on n8n, Make and Zapier, and n8n vs Zapier (/insights/n8n-vs-zapier-honest-comparison) covers the trade-offs beyond price. ### Where to start Pick the one that hurts most today. For most stores that's the stock alert or the weekly report, because both replace a manual check someone already does. Build it, run it next to the manual check for two weeks, and only switch the manual one off once the numbers match. ### Tasks, credits and executions: what one identical workflow costs on three platforms https://www.axioprax.com/insights/tasks-credits-and-executions (2026-09-22) The same intake workflow bills as 400 executions, 2,400 credits or 2,000 tasks. What each unit counts, and which plan runs out first. You have a quote from one automation platform and a quote from another, and they are not comparable. One counts tasks, one counts credits, one counts executions. Nobody will tell you what those words mean, because the answer is boring and the marketing is not. So here is the boring answer, for one workflow you probably already have: a form submission that creates a CRM record, assigns an owner, sends a notification, writes a row to a sheet, and posts to a channel. One trigger, five actions. This is the workflow from what happens to a new enquiry before anyone replies (/insights/between-the-form-and-the-first-follow-up), costed three ways. ### Why it is hard to answer The three platforms bill different things, not different amounts of the same thing. A step, a check and a whole workflow run are three different units. Comparing the numbers before you understand the units gets you a confident wrong answer. The unit changes what your architecture costs. The same decision, whether to poll or use a webhook, is free on one platform and doubles your bill on another. That is not a pricing difference. It is a counting difference, and it only shows up once you know what is being counted. The pricing pages do not answer it. All three publish prices. Only one of them lets you price this workflow end to end from the page alone. I will show you where each one stops. ### How it actually works Zapier counts successful actions. A task is any successful action that runs in Zapier. Only successful actions count toward your task usage (Zapier Help Center, task usage, checked 18 September 2026). What does not count is generous and specific: All trigger steps are free, so are any Filter or Paths step, and so are All action steps that error or halt (same page, same date). The pricing page puts it plainly: Zapier never charges you a task to check for new data (zapier.com/pricing, checked 18 September 2026). Two details that catch people. Replaying a run recharges you: Any previously successful steps that run again when you replay an entire Zap run count again (Zapier Help Center, checked 18 September 2026). And not every action costs one task. Lead Router charges 5 tasks per lead routed, Zapier MCP charges 2 per tool call, and AI and Code steps vary by model tier and runtime (same page, same date). So a task count built by multiplying steps is a floor, not a ceiling. Make counts module runs, including the checks. An operation is a single module run to process data or check for new data (Make Help Center, operations, checked 18 September 2026). The phrase or check for new data is the whole article in six words. Two rules follow. Trigger modules only run once to check for or retrieve data, regardless of the number of bundles returned, and every other module runs per bundle: modules process each bundle separately, meaning each bundle triggers its own module run (same page, same date). The pricing page uses the word credits for the same idea: Each module action in your scenario, like adding a Google Sheet row or fetching Gmail account data, counts as one credit (make.com/en/pricing, checked 18 September 2026). n8n counts whole workflow runs. An execution is a single run of your entire workflow. It doesn't matter how many steps are in the workflow or how much data it processes (n8n.io/pricing, checked 18 September 2026). The documentation adds the qualifiers that matter: Only production executions count towards this quota, schedule triggers count one execution every time the node fires, regardless of outcome, and polling nodes count one execution only when new data is found. Polls that return no results don't count as an execution (docs.n8n.io, executions, checked 18 September 2026). ### One worked calculation This is a model, not a measurement. The inputs are assumptions. Substitute your own. One trigger, five actions, 400 enquiries a month. The same workflow, counted by each platform's own rules. PlatformUnitArithmeticMonthly Zapiersuccessful actions5 x 400, trigger free2,000 tasks Make, webhook triggermodule runs6 x 4002,400 credits Make, polled every 15 minmodule runs2,880 checks + 2,000 actions4,880 credits n8n, either triggerworkflow runs1 x 400400 executions The polling row is the one to look at twice. Checking every fifteen minutes is 4 x 24 x 30 = 2,880 checks a month. On Make each is a module run, because checking for new data is explicitly a module run. On n8n an empty check is not an execution. On Zapier no check is ever a task. The same architectural decision costs 2,880 units on one platform and nothing on the other two. Now price it, using each platform's own page on the same day. - n8n: 400 executions of the 2,500 on Starter at 20 EUR a month billed annually. 16 percent used. - Make, webhook: 2,400 credits of the 10,000 shown on Core at $12 a month. 24 percent used. - Make, polled: 4,880 of the same 10,000. 49 percent used. - Zapier: 2,000 tasks is exactly the 2,000-task tier. 100 percent used, no headroom. Professional starts from $19.99 a month and the page offers tiers of 1.5K, 2K, 5K and upward, but it does not show the price at the 2K tier, so what this workflow actually costs on Zapier is UNKNOWN and I will not estimate it. What the model does not do. It assumes a flat five-action shape with no branches, no retries, no multi-task actions and no replays, all four of which push the Zapier number up and none of which push it down. The percentages use the default tier each pricing page showed on 18 September 2026, and volume selectors move them. And a plan that fits today at 400 enquiries is a different question from a plan that fits at 1,200. ### The finding, stated plainly At this volume, none of these is expensive. The difference is not the bill, it is the headroom: 16 percent, 24 to 49 percent, and 100 percent, for identical work. The platform that counts steps runs out first, and it runs out silently, because at the limit Zapier switches you to pay-per-task billing unless you have turned it off, charged at a higher per-task rate than your base subscription tasks (zapier.com/pricing, checked 18 September 2026). ### Failure modes - Comparing the numbers before the units. 2,000 against 2,400 against 400 looks like a five-fold difference. It is three different things being counted. - Budgeting from a step count. Lead Router at 5 tasks, MCP at 2, AI and Code by runtime. Your multiplication is a floor. - Polling because it was easier to build. 2,880 units a month on Make before a single enquiry arrives, free on the other two. - Paying for retries you did not know about. Replaying a run recharges every previously successful step. A week of debugging a live workflow is a real line on the bill. - Hitting the ceiling as an overage rather than an error. The workflow keeps running and the rate goes up. Nobody gets an outage, so nobody investigates. - Assuming the cheap plan is the same product. On the Make pricing page, Core at $12, Pro at $21 and Teams at $38 all showed the same 10,000 credits on 18 September 2026. You are buying features at that tier, not volume. ### When not to automate If this workflow runs a handful of times a week, none of this arithmetic matters and you should pick the platform your team can already maintain. Unit economics start mattering somewhere around the point where the monthly number stops being something you can hold in your head. And the platform question is downstream of a question this article does not answer: which steps should run unattended at all. A cheaper unit on a workflow that should have had a human in it is not a saving. ### Tradeoffs What each billing unit rewards, and what it punishes. UnitCheap whenExpensive whenThe trap Successful actions (Zapier) Few steps, high failure tolerance, no polling Many steps per item; replays; multi-task actions A step count is a floor, and the ceiling is an overage rate, not a stop Module runs (Make) Webhook triggered; few bundles per run Polled; many bundles; long scenarios Checking for data is billable even when there is no data Workflow runs (n8n) Long workflows; many steps per item Very high run counts of tiny workflows Schedule triggers count regardless of outcome, so a noisy scheduler is a real cost Self-hosting changes the n8n column entirely and is its own decision, covered separately. ### One next step Open the workflow you actually run most and count three things: how many billable steps per item, what the trigger is, and how many items a month. Multiply, then look up the allowance on your current plan and work out what percentage you are using. If it is over about sixty percent, you do not have a pricing problem yet. You have a headroom problem, which arrives first and is cheaper to fix. If you would rather have someone count it properly across your whole stack, the route is /start (/start). ### What this article is built on Twenty-one claims. Eighteen verified against primary vendor pages read on 18 September 2026, three published as unknown rather than guessed. Each is cited in the text at the point it is used, with the date it was checked. Claim ledger. Absent data is recorded as unknown, never as zero. #ClaimSourceCheckedStatus 1"A task is any successful action that runs in Zapier. Only successful actions count toward your task usage."Zapier Help Center, task usage18 Sep 2026Verified 2"All trigger steps" do not count as taskssame18 Sep 2026Verified 3"any Filter or Paths step" does not count as a tasksame18 Sep 2026Verified 4"All action steps that error or halt" do not countsame18 Sep 2026Verified 5Replayed previously successful steps are charged againsame18 Sep 2026Verified 6Some actions cost more than one task: Lead Router 5, Zapier MCP 2, AI and Code variablesame18 Sep 2026Verified 7"Zapier never charges you a task to check for new data"zapier.com/pricing18 Sep 2026Verified 8Free $0 a month with 100 tasks; Professional from $19.99 a month; Team from $69 a month; tiers include 1.5K, 2K, 5K and upzapier.com/pricing18 Sep 2026Verified 9At the limit, pay-per-task billing applies unless disabled, at a higher ratezapier.com/pricing18 Sep 2026Verified 10"an operation is a single module run to process data or check for new data"Make Help Center, operations18 Sep 2026Verified 11Trigger modules run once "regardless of the number of bundles returned"same18 Sep 2026Verified 12"modules process each bundle separately, meaning each bundle triggers its own module run"same18 Sep 2026Verified 13"Each module action in your scenario ... counts as one credit"make.com/en/pricing18 Sep 2026Verified 14Free $0 up to 1,000 credits; Core $12, Pro $21, Teams $38 each shown at 10,000 credits; Enterprise custommake.com/en/pricing18 Sep 2026Verified 15"An execution is a single run of your entire workflow. It doesn't matter how many steps are in the workflow"n8n.io/pricing18 Sep 2026Verified 16Starter 20 EUR a month 2.5K executions; Pro 50 EUR a month 10K; Business 667 EUR a month 40K, billed annuallyn8n.io/pricing18 Sep 2026Verified 17"Only production executions count towards this quota"; schedule triggers count "regardless of outcome"docs.n8n.io, executions18 Sep 2026Verified 18Polling nodes "count one execution only when new data is found. Polls that return no results don't count as an execution."docs.n8n.io, executions18 Sep 2026Verified 19What 2,000 Zapier tasks a month actually coststhe pricing page shows the 2K tier but not its price18 Sep 2026UNKNOWN 20Whether an n8n execution that fails after starting counts toward quotathe docs address pre-start failures and schedule triggers, not mid-run failure18 Sep 2026UNKNOWN 21Whether Make's "operations" and "credits" are strictly the same unitthe pricing page says credits, the operations help page says operations, neither reconciles them18 Sep 2026UNKNOWN Third-party posts claiming a September 2026 Make price cut were surfaced during research and deliberately not admitted. An aggregator blog is not vendor documentation, and every other claim here was held to that bar. ### CRM Automation for Small Businesses: The First Five to Build https://www.axioprax.com/insights/crm-automation-small-business-guide (2026-08-27) What CRM automation actually means, the five automations we would build first, what HubSpot, Salesforce, Zoho and Pipedrive charge before you can automate, and the mistakes to skip. Most small businesses buy a CRM, import their contacts, and then use it as a slightly more expensive spreadsheet. Nobody updates the deal stages, follow-ups live in someone's head, and six months later the data is too messy to trust. Automation is what turns a CRM from a list into something that does work for you. This guide covers what CRM automation actually is, the first five automations we'd build in almost any small business, what HubSpot, Salesforce, Zoho and Pipedrive charge before you can automate anything, and the mistakes that make it backfire. Prices were read on each vendor's own pricing page on 20 October 2026. ### What CRM automation actually means Strip away the jargon and it's this: when something happens in your CRM, something else happens without anyone having to remember. It comes in three kinds. - Data. Records fill themselves in. A form submission becomes a contact, an order becomes a purchase on that contact, a company name gets matched to the right account. - Process. Work moves along. A new lead gets an owner, a deal that hasn't moved in a week gets flagged, a won deal creates the onboarding tasks. - Communication. The right message goes out at the right time. A welcome email, a reminder before a quote expires, a check-in after the first month. Most businesses start with communication because it's the most visible. We'd start with data. If the records are wrong, every email and every report built on them is wrong too. ### The first five automations we'd build These five cover most of what a small sales team loses time on. Build them in this order. - Form to contact. Every enquiry from your website lands in the CRM as a complete contact, with the source recorded. Nobody retypes anything. - Owner assignment. Each new lead gets an owner straight away, by round robin, territory or deal size. An unowned lead is a lead nobody calls. - Follow-up task. The owner gets a task with a due date the moment the lead arrives. This one matters most. We wrote a whole piece on what happens to an enquiry before anyone replies (/insights/between-the-form-and-the-first-follow-up). - Stalled deal nudge. If a deal sits in the same stage for a set number of days, the owner hears about it. Deals rarely die loudly. They go quiet. - Orders into the CRM. If you sell online, every order should show up against the customer, so sales knows who bought what before picking up the phone. On that last one: if you run Shopify and use HubSpot, the native integration is free on all HubSpot plans. According to HubSpot's own documentation (https://knowledge.hubspot.com/integrations/connect-hubspot-and-shopify-data-sync), contacts and products sync both ways, while orders and abandoned carts come across one way, from Shopify into HubSpot, as orders rather than deals. If you want HubSpot deals to push anything back to Shopify, that isn't part of the native integration. You'd need another app or a custom build, which is what our Shopify to HubSpot sync (/work/shopify-hubspot-sync) demo does. ### What each CRM charges before you can automate This is where people get caught out. The free or entry plan is often enough to store contacts but not to automate them. CRMEntry priceWhere automation starts HubSpotFree for up to 2 users. Starter lists at $20 per seat a month (a $7 offer for new customers is a limited-time promotion)Free has no workflows. Starter allows up to 50 workflows with restricted triggers. Professional, from $90 to $100 per seat plus a $1,500 onboarding fee, is the first tier with fully customisable workflows SalesforceStarter Suite, $25 per user a monthStarter includes built-in sales flows and lead routing. Pro Suite, $100 per user a month billed annually, adds more customisation and automation Zoho CRMFree for up to 3 usersWorkflow automation is included on the free plan PipedriveLite, $14 per seat a month billed annuallyAutomations start on Growth, $39 per seat a month billed annually Sources: HubSpot (https://www.hubspot.com/pricing/sales), Salesforce (https://www.salesforce.com/small-business/pricing/), Zoho (https://www.zoho.com/crm/zohocrm-pricing.html), Pipedrive (https://www.pipedrive.com/en/pricing). ### HubSpot or Salesforce? This is the question we get asked most, so here's our honest take. HubSpot is the easier place to start. The free CRM is genuinely usable, the interface is friendlier for a team without an administrator, and most small businesses can set up the five automations above on Starter or Professional without outside help. If you would rather hand it over, that is our CRM automation setup (/services/crm-automation). Salesforce gives you more control over how records, fields and flows behave, and it scales into very complex sales processes. That flexibility comes with more setup and more ongoing admin, and Pro Suite needs an annual contract. If you have a large or unusual sales process, or a team that will grow quickly, it's worth the extra effort. If you don't, it's usually more CRM than you need. If you're already on Salesforce, the same five automations apply. Salesforce Flow can route leads by territory or size, create follow-up tasks when a deal changes stage, warn reps before a quote expires, kick off onboarding when a deal closes, and alert account owners when a customer goes quiet. ### Mistakes that make CRM automation backfire - Automating messy data. If half your contacts have no email or the wrong company, automation just spreads the mess faster. Clean first. - Building everything at once. Ten automations launched on the same day are impossible to debug. Add one, watch it for a week, then add the next. - No owner. Someone has to be responsible when an automation misfires. If nobody is, it'll quietly fail for months. - Emails that fire on stale information. A "thanks for your order" email to someone who cancelled yesterday does more harm than no email. - Never testing with real records. Test data is tidy. Real data has duplicates, blank fields and odd characters. Test with a handful of real contacts before switching anything on. ### Where to start Start with whatever is leaking money right now. In most small businesses that's follow-up: enquiries that arrive and wait too long for a reply. Get form to contact, owner assignment and the follow-up task working, and you'll feel the difference within a week. Everything else builds on those three. ### Between the form and the first follow up: what happens to a new enquiry before anyone replies https://www.axioprax.com/insights/between-the-form-and-the-first-follow-up (2026-09-18) A form submission succeeds silently. What actually happens to a new enquiry in HubSpot and Shopify before a human sees it, and where ownership goes missing. Someone filled in your form on Friday evening. Nobody replied until Monday. The uncomfortable part is not the delay. It is that when you ask whose job that reply was, nobody can answer, and there is no place in the system where the answer was ever recorded. Most advice here is about speed, and you already believe speed matters. What you cannot see is the stretch between a form submission and a human reading it, because every step in it succeeds quietly. This article walks that stretch in the two stacks a one to fifty person company usually has, and shows where ownership goes missing. ### Why it is hard to answer The steps are invisible because none of them fail. The form submits. The record is created. The email is sent, or is not, according to a setting nobody has looked at since setup. Nothing raises an error, so nothing tells you an enquiry was captured correctly and then read by no one. Ownership is a setting, not a conversation. HubSpot's documentation is direct about the consequence: If a contact has no owner, no notifications will be sent (HubSpot Knowledge Base, form submission notifications, checked 18 September 2026). A correctly captured enquiry can be completely silent, and the silence looks identical to a quiet week. The fix you assume is paywalled is mostly not. Both the record ownership page and the form notification page carry the requirement line All products and plans (HubSpot Knowledge Base, checked 18 September 2026). Setting an owner and sending notifications are available across plans. Only automatic distribution is gated, which is a narrower problem than most people expect. ### How it actually works Start with the asymmetry, because everything else follows from it. If a user manually creates a record, that user will be automatically set as the record's owner, unless they select a different owner during creation (HubSpot Knowledge Base, record ownership, checked 18 September 2026). When a person types a contact into the CRM, ownership is automatic. When a form creates that same contact, nobody is named. The system behaves differently depending on who created the record, and the case with no human in the loop is exactly the case that gets no owner. The HubSpot path. A submission arrives. If the email matches an existing contact, the details on the existing record will be updated (HubSpot Knowledge Base, create and edit forms, checked 18 September 2026); otherwise a new contact is created. Form-level automation then runs inside tier limits: one action on free accounts, three on Starter, Professional or Enterprise without Marketing Hub, ten on Marketing Hub Starter, unlimited with multiple workflows on Marketing Hub Professional and Enterprise (HubSpot Knowledge Base, form automations, checked 18 September 2026). An owner is assigned, or is not. Notifications fire, or do not. And the recipient list has its own quiet rule: Notification recipients must be active users in your HubSpot account (HubSpot Knowledge Base, checked 18 September 2026). The tier wall sits on distribution. The assignment rotation workflow action requires a Sales or Service Hub Professional or Enterprise subscription plus an assigned seat (HubSpot Knowledge Base, record ownership, checked 18 September 2026). The Shopify path. The default theme contact form is thinner. Your contact form sends all submissions to your store's sender email address (Shopify Help Center, add a contact page, checked 18 September 2026). That is all the page documents. Whether the submission creates any customer record is not stated on it; that page has now been read five times across seven days and is silent every time. Silence is not denial, so this article will not tell you no record is created. It tells you the documentation does not describe one, and that this is a two minute test in your own store. The separate Shopify Forms app behaves differently: A customer profile is created for new customers, and existing customers have their information matched to their profile, though Contact information isn't updated when an existing customer submits a form (Shopify Help Center, manage forms, checked 18 September 2026). Do not read that across to the theme contact form. The common point of failure in both stacks is the handoff from captured to owned. It is the only step in the chain with no system of record. ### One worked calculation This is a model, not a measurement. The inputs are assumptions. Substitute your own. Assume 30 web enquiries a month, working hours Monday to Friday 9am to 5:30pm, and for the moment that enquiries arrive evenly across the week. Working hours per week: 5 x 8.5 = 42.5. That leaves 168 - 42.5 = 125.5 hours, so 125.5 / 168 = 74.7 percent of the week is outside working hours. Applied to 30 enquiries: 30 x 0.747 = 22.4 enquiries a month arriving when nobody is working. The even-arrival assumption is wrong in a known direction: B2B enquiries cluster in working hours, so 74.7 percent is an upper bound, not an estimate. What survives the objection is the single case. From 5:30pm Friday to 9am Monday is 63.5 hours, and an enquiry arriving then with no owner and no out-of-hours rule waits all of it before anyone is scheduled to look. For scale, an audit of 2,241 US companies found that among those that responded at all within 30 days, average first response time was 42 hours, and that 37 percent replied within an hour while 23 percent never replied (Oldroyd, McElheran and Elkington, The Short Life of Online Sales Leads, Harvard Business Review, March 2011). That study is fifteen years old and US-based. A single unowned weekend enquiry exceeds its 42 hour average on its own. The same study reports firms contacting a lead within an hour were roughly seven times as likely to qualify it, and more than sixty times as likely as those waiting a day or more. This model does not multiply anything by sixty. That figure is an association across a sample whose base rates are not yours, and turning it into revenue would produce a large number that means nothing. The model stops at hours waited. A current benchmark would be better, but there is no credible one. The most prominent reachable 2026 report calls itself a compilation and traces its headline figures to that same 2011 dataset, with no original fieldwork (checked 13 September 2026). Whether response times have improved since is unknown. ### Failure modes, and the symptom of each - Unowned contact. No notification fires. Symptom: enquiries discovered later in the CRM, already cold. - Notification to a shared inbox with no named owner. Symptom: replies that begin "I thought you had this one." - Out-of-hours arrival. Not lost, queued behind a weekend. Symptom: your slowest replies cluster on Mondays. - Duplicate under a second email address. Matching is on email, so a second address makes a second record. Symptom: two half-conversations with one person. - Automation built, working, with nobody owning its exception queue. A failed action is silent in exactly the way the original problem was silent. - A departed colleague still on the notification list. Recipients must be active users, so a deactivated one drops off quietly. Symptom: a route that worked last quarter and now goes nowhere. ### When not to automate Below roughly five enquiries a week, a named person and a recurring calendar reminder beat a workflow, which becomes one more thing to maintain that the volume never repays. Automate when volume is steady, when enquiry types are similar enough to route by rule, and when someone owns the exceptions. Do not automate to compensate for nobody owning intake. That is the problem, and a workflow on top of it just moves the silence somewhere harder to see. ### Tradeoffs Six ways to handle intake ownership, and where each one stops working. Approach What it costs What it fixes What it breaks When it is the wrong choice Shared inbox, no assignment NothingNothing. This is the baseline. Ownership is ambiguous by defaultAny volume above a handful a week One named owner for all intake Agreement, nothing moreAmbiguity, immediately Single point of failure, holidaysWhen that person is also delivery Manual triage rota A recurring calendar commitmentOwnership and out-of-hours cover Depends on discipline, fails quietlyWhen the rota has no visible record CRM auto-assign to one owner Available on all plansNotification silence Everything lands on one personWhen enquiry types need different people HubSpot assignment rotation Sales or Service Hub Professional or Enterprise, plus a seat Distribution and ownership together Cost step up, seats to manageWhen the team is one or two people Custom router, webhook to n8n or similar Build and maintenance timeRouting by type, out-of-hours rules A new failure surface needing its own monitoringWhen nobody will own the exception queue ### One next step Open your own account and answer three questions. - When a form is submitted, does the record get an owner, or does it sit unassigned? - Who is named, by name, as responsible for the first reply? - What happens to an enquiry that arrives at 6pm on Friday? If the first two have answers and the third has a rule, your intake is fine and you do not need to automate anything. If one of them does not, that gap is where your Friday enquiry went, and it is worth fixing before you buy a tool to sit on top of it. If you would rather have someone map that stretch for you, the route is /start (/start). ### What this article is built on Fifteen claims sit behind this article. Twelve are verified against primary vendor documentation, and three are stated as unknown rather than guessed. Every vendor claim was re-read against its own page on 18 September 2026, and each is cited in the text at the point it is used, with the date it was checked. Claim ledger. Absent data is recorded as unknown, never as zero. #ClaimSourceCheckedStatus 12,241 US companies audited; 37% replied within an hour, 23% neverHarvard Business Review, March 201112 Sep 2026Verified 2Average first response 42 hours among those responding within 30 daysSame as 112 Sep 2026Verified 3Roughly 7x and more than 60x qualification likelihoodSame as 112 Sep 2026Verified, association only, deliberately not multiplied 4Email match updates the existing record; no owner means no notificationsHubSpot Knowledge Base, create and edit forms; form submission notifications18 Sep 2026Verified 5Assignment rotation needs Sales or Service Hub Professional or Enterprise plus a seatHubSpot Knowledge Base, record ownership18 Sep 2026Verified 6Form automation tier limits of 1, 3, 10 and unlimited actions; webhooks are an action typeHubSpot Knowledge Base, form automations18 Sep 2026Verified 7Theme contact form sends submissions to the store's sender email addressShopify Help Center, add a contact page18 Sep 2026Verified 8Whether a theme contact form creates any customer recordSame page, silent on the question18 Sep 2026, fifth readUNKNOWN 9A current 2025 or 2026 primary benchmark for SMB response timeNone found that is not a compilation of the 2011 data18 Sep 2026UNKNOWN 10Share of UK SMB enquiries arriving out of hoursNo primary source; the calendar model stands in12 Sep 2026UNKNOWN 11Shopify Forms app creates a customer profile; existing contact information is not updatedShopify Help Center, manage forms18 Sep 2026Verified, not merged with claim 8 12The most prominent 2026 speed-to-lead report is a compilation of the 2011 datasetPublisher's own methodology statement13 Sep 2026Verified, cited only to explain declining a benchmark 13Notification recipients must be active usersHubSpot Knowledge Base, form submission notifications18 Sep 2026Verified 14Manually created records are owned by their creator automaticallyHubSpot Knowledge Base, record ownership18 Sep 2026Verified 15Both HubSpot pages carry the line "All products and plans"; only rotation is tier gatedBoth pages above18 Sep 2026Verified ### Stocky is gone: how do I rebuild Shopify reorders? https://www.axioprax.com/insights/stocky-is-gone-rebuild-shopify-reorders (2026-11-02) Stocky closed on 31 August 2026. What Shopify's purchase orders, suppliers, stock counts, reports and Sidekick cover now, the seven gaps that remain, and three ways to rebuild reordering. Stocky used to answer three questions when you raised a purchase order: what to order, from whom, and how many. That screen is gone. Shopify says Stocky is no longer available as of 31 August 2026, its APIs stopped working that day, and merchants have read-only access to export their data for at least 90 days after (Shopify Help Center, migrating from Stocky (https://help.shopify.com/en/manual/inventory-and-locations/transitioning-from-stocky)). Shopify gives no end date, so the safe reading is that the export window closes no earlier than about 29 November 2026. This piece maps what Stocky did for reordering, what Shopify gives you in the admin today, what is still missing, and three ways to rebuild the routine. Where we tested something, it was on Northfield Outfitters, our own Shopify development store, with a fictional shop and synthetic data. There are no client results here. ### What exactly did Stocky do for reordering? Stocky forecast the stock you needed and suggested which products to order. Shopify describes it as an inventory app included with Shopify POS Pro subscriptions that helped by forecasting the inventory that you need, and suggesting which products you should order (Shopify Help Center, Stocky (https://help.shopify.com/en/manual/sell-in-person/shopify-pos/inventory-management/stocky)). The maths was simple, and Shopify documents it. The default method, "Last X days", takes average orders per day over a sample period and multiplies it by the number you type into "Suggest for (days)". Shopify's own example: 14 shirts sold in the last 7 days is 2 a day, and a 14-day forecast suggests ordering 28 (Shopify Help Center, demand forecasting in Stocky (https://help.shopify.com/en/manual/sell-in-person/shopify-pos/inventory-management/stocky/inventory-management/demand-forecasting)). There were other bases too: - Custom date range: the same daily average, taken from dates you pick, such as a past sale. - Same period last year: order what sold in the same window last year. - Fill shelves: order up to the shelf limits set on the supplier or vendor page. - Target stock level: order up to one number. With stock of 3, 6 and 10 and a target of 15, Stocky suggests 12, 9 and 5. Once a purchase order was created, you could send it to the supplier from inside Stocky. None of this needs special software. It needs daily sales, current stock and a rule. That is why it can be rebuilt. ### What are the dates that still matter? Two dates matter now: 31 August 2026, when Stocky closed and its APIs stopped, and about 29 November 2026, the earliest the read-only export can close. Shopify's wording is "at least 90 days", so the window may stay open longer, but nothing promises that. Treat late November as the deadline. Shopify says to export three things using Stocky's reports: all completed purchase order reports as CSV files, your stocktake history and your historical cost data. Suppliers can't be exported from Stocky, so screenshot every supplier record. Historical purchase orders can't be imported into Shopify either: the native CSV upload only adds product lines to a new draft purchase order. Keep the files in a dated folder you own. Our free checklist for old Stocky data (/store-kits/stocky-rescue#export-checklist) lists every file and screenshot in order. It is a one-page PDF and needs no email. ### What does Shopify give you instead? Shopify gives you the paperwork: native purchase orders, suppliers, receiving through transfers, stock counts, inventory reports and, since October 2026, purchase order reports, plus Sidekick and Flow for suggestions and alerts. Shopify announced native purchase orders on 1 May 2023 (Shopify Changelog (https://changelog.shopify.com/posts/use-purchase-orders-to-order-and-stock-inventory-from-your-suppliers)). ### Can Shopify do purchase orders on its own now? Yes. Go to Products > Purchase orders. A purchase order has two statuses, Draft and Ordered, and marking it Ordered can't be reversed. If you have ordered a product from that supplier before, Shopify fills in the Supplier SKU, Cost and Tax from your previous purchase orders (Shopify Help Center, creating purchase orders (https://help.shopify.com/en/manual/inventory-and-locations/purchase-orders/creating-purchase-orders)). Stock is received on the inventory transfer linked to the purchase order, not on the purchase order itself. When every unit is accepted, rejected or cancelled, the transfer shows Transferred. Shopify also now says previously closed purchase orders are displayed as archived: they appear greyed out in the purchase order list, and the order's own page shows an Archived badge. On our demo store, between 16 and 19 October 2026, we raised purchase order #PO2 for $7,550.00, marked it Ordered and received it through transfer #T0002, including one board that arrived late. ### Where do suppliers, counts and reports live? Suppliers are created from the purchase order screen and hold contact details and an address. Payment terms and supplier currency are set on each purchase order, not saved on the supplier, and payments to suppliers happen outside the Shopify admin (Shopify Help Center, managing suppliers (https://help.shopify.com/en/manual/inventory-and-locations/purchase-orders/managing-suppliers)). Stock counts live under Inventory > Stock counts. Staff count, and a manager can review and approve before stock levels change (Shopify Help Center, stock counts (https://help.shopify.com/en/manual/products/inventory/inventory-counts)). For demand, the Inventory reports (Analytics > Reports, then filter Category to Inventory) are the closest thing to Stocky's numbers. "Inventory remaining per product" divides stock by average daily sales over the last 28 days, and "Inventory sold daily by product" gives the daily rate itself. ABC analysis grades products by revenue, but stores on Basic or Lite need an app for it. One change to note: from 1 October 2026, inventory reports measure On hand instead of Available, so recent figures can read higher than older ones (Shopify Help Center, inventory reports (https://help.shopify.com/en/manual/reports-and-analytics/shopify-reports/report-types/default-reports/inventory-reports)). For spending, Shopify added three purchase order reports to Analytics on 6 October 2026: Purchase order costs, which covers ordered items, quantities, supplier costs, estimated taxes and total cost; Purchase order transfers; and Purchases on order (Shopify Changelog (https://changelog.shopify.com/posts/analyze-purchase-orders-in-shopify-analytics)). ### Does anything in Shopify suggest what to reorder? Shopify points to Sidekick. Its migration page gives "What should I reorder?" as an example question and says Sidekick recommends items based on sales data and can draft the purchase order for you. Shopify Flow, Shopify's free automation app, has two relevant templates: "Get notified by email when product variant inventory is low" and "Initiate reorder with vendor when inventory gets low". Both start from the trigger that fires when a variant's inventory changes (Shopify Help Center, Flow trigger (https://help.shopify.com/en/manual/shopify-flow/reference/triggers/product-variant-inventory-quantity-changed)). Our guide to Shopify automation workflows (/insights/shopify-automation-workflows-guide) covers other routine store jobs. Our honest limit here: we have not tested Sidekick's reorder suggestions on our demo store, and Sidekick's own help page does not explain how it forecasts, which plans get it, or whether it uses lead times. Check its suggestions against your own lead times before you trust them. ### What is still missing compared with Stocky? Deciding what to order is the missing part. Shopify gives you the documents, but the decision rules Stocky stored (lead times, minimums, shelf limits) have no working home in the admin. Shopify's own migration page lists the gaps: - Sending a PO. Stocky could send it from inside the app. In Shopify you can't email a purchase order from the admin; you export a PDF and send it yourself. - Customer orders and merging. You can't turn a customer order into a purchase order or merge purchase orders. - Spend by supplier. Shopify's migration page still says there isn't a dedicated native report for total purchase order spend by supplier. The new Purchase order costs report includes supplier costs, so check whether it gives you the rollup you need before you build one. - Lead times, minimum orders and case packs. These can live in variant metafields, but they are for reference only and don't feed purchase order calculations. - Minimum and maximum levels. Shopify suggests custom metafield definitions, reports, or Sidekick. - Old purchase orders. They can't be imported. - Costing. Each variant has one static Cost per item; weighted-average or landed cost needs a third-party app. Two more things caught us out on the demo store. Products with inventory tracking off were greyed out and labelled "Inventory not tracked" when we added them to a purchase order, which fits Shopify's instruction to activate tracking before you create one. And the last cost filled itself in: #PO2 opened with $700 per board carried over from an earlier test order, so we retyped every line. ### What are the three ways to rebuild reordering? You can stay native, add a reorder sheet to native purchase orders, or move to a third-party app. The right one depends on how many suppliers and locations you run. ### Path 1: native only (Sidekick, Flow and reports) This suits a small catalogue with a handful of suppliers. The steps: - Turn on inventory tracking for every product you reorder. - Install the Flow low-stock email template and set your threshold. Our fix page walks through it: get a low stock alert without an app (/fixes/shopify-low-stock-alert-without-app). More inventory fixes sit in the Inventory area of Fixes (/fixes). - Each week, open "Inventory remaining per product" and ask Sidekick what to reorder. - Raise a native purchase order, export the PDF and send it. The weakness: none of these steps holds a supplier's lead time or minimum order in the maths, and the method lives in one person's head. ### Path 2: native purchase orders plus a reorder sheet This is Stocky's "Last X days" rule with lead time added. In our reorder sheet, each product's reorder point comes from its daily sales, its supplier's lead time and a safety stock. A simple version is average daily sales multiplied by lead time plus safety days. An example with round numbers: a product selling 2 a day, with a 10-day lead time and 4 safety days, has a reorder point of 28. At or below 28 in stock, it goes on this week's order. This is what our Stocky Replacement kit (/store-kits/stocky-rescue) does. Your suppliers go into Shopify's own purchase orders, a supplier sheet keeps what Shopify doesn't hold (lead times, minimum orders and payment terms), a reorder sheet and a Monday email decide what to order, and your first stock count runs in Shopify. On our demo store, the reorder sheet first held 10 of 18 products out of the order for missing, duplicate or absent sales data. Once we filled the gaps, the products marked ORDER NOW went from 2 to 4, and the Monday email's $7,550 order to one supplier matched purchase order #PO2 to the cent. The full build, including the test rounds it failed, is in our demo build write-up (/insights/stocky-rescue-demo-build). The weakness: it is a weekly step, it has no seasonality, and a sheet needs an owner. On the demo store the Monday email has run once by hand from a real Google account; a run on its weekly schedule is not yet proven. ### Path 3: a third-party app This suits several warehouses, landed costs, manufacturing or many suppliers. Shopify itself points to third-party inventory apps for landed costs and bin-level moves. Some App Store listings, in their own words, with no ranking from us and no testing by us: - EZstock (https://apps.shopify.com/ezstock): "See what to reorder and when, email purchase orders... Leaving Stocky? Upload your export." - Binly (https://apps.shopify.com/binly): "Stocky is gone. Binly replaces it" with "phone barcode stocktakes, purchase orders, smart reorder". - Stockcast (https://apps.shopify.com/stockcast-inventory-forecast): "Purchase orders by supplier and Stocky CSV import." - Prediko (https://apps.shopify.com/prediko): "Demand forecasting, inventory transfers, raw materials, manage purchase orders & reports." - Inventory Planner by Sage (https://apps.shopify.com/inventory-planner): "Forecast demand, optimize inventory & streamline purchases". - Stockroom by MyWorks (https://apps.shopify.com/stock-room): "Purchase orders, inventory, manufacturing with BOMs, and QuickBooks sync". Several of these pitch themselves as Stocky replacements, so check each listing's launch date and review history before you commit. Ask each one two questions: does it import a Stocky purchase order export, and how far back does it read your orders? ### How do you choose the right path for your store? Pick by supplier count, locations and how much of Stocky's logic you used. - Few suppliers, one location, and you mostly reorder from memory: Path 1. Write each supplier's lead time down anyway. - You liked Stocky's suggestions and want them back with lead times and minimums: Path 2. - Several warehouses, landed costs, manufacturing or dozens of suppliers: Path 3. Whatever you pick, two rules hold. Shopify only treats an item as out of stock when inventory is tracked, and untracked products would not go on our purchase orders, so switch tracking on first. And don't let the app decision delay the export. ### What should you do this week? Get the data out first, then decide. - Export completed purchase order reports, stocktake history and historical cost data from Stocky before about 29 November 2026. - Screenshot every supplier record. - Write each supplier's lead time, minimum order and payment terms on one page. - Raise your next order as a native Shopify purchase order and receive it through its transfer. If you want the Monday routine rebuilt for you, see the Stocky Replacement kit (/store-kits/stocky-rescue), then describe your store (/start?kit=stocky-rescue). ### Can AI agents buy from your Shopify store? https://www.axioprax.com/insights/can-ai-agents-buy-from-your-shopify-store (2026-11-02) Which AI channels can sell your Shopify products today, which take the order inside the chat, where the settings sit, what the agents read about your products, and what to fix this week. It is easy to assume AI shopping is something you sign up for later. On Shopify it is already running. Shopify's help page says agentic storefronts let customers discover and purchase your products in AI channels and that the feature is active by default for eligible stores (Shopify Help Center, agentic storefronts (https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts)). This piece covers what is live today, channel by channel, where the settings sit, what the agents actually read about your products, who owns the order, and the short list we would work through this week. Everything here comes from Shopify and OpenAI pages we opened on 31 October 2026 and checked again on 2 November. One honest limit up front: we have not yet placed an agent order on our own demo store, so there are no test results in this article, only what the platforms document. ### Which AI channels can take an order today? Shopify's help page names four third-party AI channels, and they work in two different ways. Shopify's own Shop app is a fifth agentic storefront, covered below. - ChatGPT is a discovery channel. Shoppers find your products in a chat, then pay on your normal online store checkout, opened in ChatGPT's in-app browser or a new tab. Shopify says there is no option to activate or deactivate a ChatGPT direct checkout setting (Shopify Help Center, selling on ChatGPT (https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/chatgpt)). OpenAI says shoppers can pay on your website with ChatGPT Wallet, and your checkout processes the payment (OpenAI, accept payments (https://developers.openai.com/commerce/guides/key-concepts)). - Google AI Mode and Gemini have direct checkout. The shopper clicks Buy, fills in shipping and card details, and pays without visiting your store (Shopify Help Center, selling on Google AI Mode and Gemini (https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/google)). - Microsoft Copilot has direct checkout for US shoppers paying by credit or debit card (Shopify Help Center, selling on Microsoft Copilot (https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/microsoft)). - Meta surfaces such as Muse have direct checkout too (Shopify Help Center, selling on Meta (https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/meta)). There is one more route that gets less attention. Shopify's help page says customers can also pay with Shop Pay when they shop through personal AI agents, such as Muse and other agents they connect to Shop, and those orders show in your admin with attribution to the agent. You don't need to switch anything on for that. Shopify's 28 October 2026 announcement still describes Shop Pay checkout in personal AI agents as coming, so expect it to reach shoppers gradually (Shopify Fall '26 Edition (https://www.shopify.com/news/fall-26-edition-merchant)). ### Which stores are eligible, and where? Eligibility depends on who you sell to, not where your business is based. That matters if you run your store from outside the US. - ChatGPT: you must sell to customers in the United States, but your store can be based outside the United States. - Google AI Mode and Gemini: you must sell to customers in the United States, Canada or Australia, and your store can be based in any country. You also need a valid Google Merchant Center account with your products in it. - Microsoft Copilot: stores selling to US customers. - Meta: stores selling to the United States, Canada, Mexico, Brazil, Japan or Singapore. Each channel's page also asks for three completed policies in Settings, then Policies: Terms of service, Privacy policy, and Return and refund policy. And Google's direct checkout is still "rolling out gradually" to shoppers in those three countries, so a shopper without it is sent to your store to finish the purchase. Shopify Catalog has its own account checks. Your store must be on the Starter plan or higher and not password protected, and Shopify reviews account standing over time: a verified email, two-step authentication, a history of real sales, prompt fulfillment and a low chargeback rate. If your store is not eligible, the Agentic page may show the reason next to Sources, then Shopify Catalog (Shopify Help Center, Shopify Catalog requirements (https://help.shopify.com/en/manual/shopify-catalog/requirements)). ### Where does this live in your admin? Everything sits in Sales channels, then Agentic. The default setting is "Allow Shopify to manage for me". With it on, AI channels can read your products through Shopify Catalog, direct checkout is on for the channels that offer it, and you are auto-enrolled in new agentic storefronts as they launch (Shopify Help Center, managing agentic storefronts (https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/agentic-home)). From the same page you can see how many of your products are in Shopify Catalog, preview how they show in Catalog search, read sales, orders, sessions and conversion for each AI channel, and open your knowledge base. If you are on Shopify's newer Agentic plan, the setup is different and direct checkout is off by default. ### What do the AI agents actually read about your products? Mostly Shopify Catalog, which is built from your product data. For Google AI Mode and Gemini, discovery comes from Google Merchant Center instead, fed either by the Google & YouTube sales channel, with automatic product and shipping sync turned on, or by a third-party feed provider or manual upload. To be included in Shopify Catalog automatically, a product needs: - A title and at least one product image. - A price above 0. Free products are left out. - To be published to your online store, or a Hydrogen or Headless storefront, with an identifiable product URL. - Not to be Unlisted or hidden from search engines. - Not to fall in an excluded category, such as alcohol, tobacco and vaping, weapons, services and add-ons, event tickets, or products Shopify classifies as Uncategorized. That last point catches people out. Shopify classifies products from your product information, and that classification can differ from the product category that you set in your admin. Thin titles and descriptions make a product more likely to be misread. Inside the Agentic page, the search preview lets you enter a product name or keyword and see the raw Catalog results. Each product gets a Listing quality meter built from five signals: description completeness, image coverage, product reviews, variant and option completeness, and shop policy completeness. Shopify adds that the variant check looks for option names with acronyms or numbers that are hard for agents to understand. So an option named "Colour" with values like "Navy" reads better to an agent than one called "Opt 2". Two more details from Shopify's channel pages: put any legal disclosures in the first 6,000 characters of the product description, and B2B-only products are not supported. For questions that are not about one product, such as shipping times or returns, there is a knowledge base where you write the facts agents can use to answer. Since 26 October 2026 it is built into the Shopify admin, where Shopify drafts those facts from your storefront for you to review and publish (Shopify Changelog (https://changelog.shopify.com/)). ### Who owns the order, and what changes at checkout? You do. Shopify's terms say you remain the seller of record (Shopify Agentic Storefronts terms (https://shopify.com/legal/terms-agentic-storefronts)), and its help page says you retain full ownership of the customer relationship and post-purchase experience. Orders from AI channels show in your admin with channel or referrer attribution. Direct checkout is not a copy of your own checkout. Shopify documents the limits in most detail for Google AI Mode and Gemini, and Copilot and Meta have their own lists (Meta's page says checkout blocks aren't supported there at all). In Google AI Mode and Gemini: - Google Analytics and custom pixels don't fire. Only server-to-server pixels for checkout started and completed run. - Some checkout blocks, such as upsells, custom fields, loyalty, address autocomplete and trust badges, might not show. - Subscription-only products, bundles, customisable products and B2B-only products can't be bought there. - Local delivery, pickup in store and pickup points aren't offered. - You can't require customers to sign in before checkout. What does work: Shopify Functions such as shipping, discount and checkout validation functions, plus your automatic discounts and discount codes. ### What does it cost? For ChatGPT, Google AI Mode and Gemini direct checkout, Microsoft Copilot direct checkout and Meta's direct checkout, Shopify's help pages say there are no channel fees, only your standard payment processing fees. Shopify's Agentic Storefronts terms say you pay the fees for each channel that are set out in your admin, so treat the admin as the final word if it ever shows a fee. ### How do you turn it off if you don't want it? Go to Sales channels, then Agentic, and turn off "Allow Shopify to manage for me". You can then: - Turn off direct checkout for one channel, so shoppers still find you but pay on your store. - Turn off Shopify Catalog access for one channel, or for all third-party channels. Shopify says it can take up to 7 days before your data stops being shared, and turning Catalog access off also turns off direct checkout. - Turn off "Auto enroll new storefronts" so new AI channels are not added for you. Two limits. Turning off Catalog access here does not apply to Shop, Shopify's own app, which is managed from its own sales channel settings, and it does not stop AI tools finding your public pages through web crawling. The only way to hide a product from every agentic storefront, Shop included, is the Unlisted status or the seo.hidden metafield, which also hides it from search engines like Google. ### What should a store owner do this week? Five jobs, in this order: - Open Sales channels, then Agentic. Check you are eligible, which channels are on, and how many products are in Shopify Catalog. - Complete your Terms of service, Privacy policy, and Return and refund policy in Settings, then Policies. - On your top 20 products, tighten titles and descriptions, add images, set clear variant option names, and move any legal disclosure into the first 6,000 characters. - Review the facts in your knowledge base, especially shipping and returns, then try three product names or keywords a shopper might use in the search preview. - If you sell to the US, Canada or Australia, connect Google Merchant Center and check whether you sell subscription-only products, bundles or customisable products, or offer pickup, because Google's direct checkout does not support them. If you rely on Google Analytics or client-side pixels to count sales, expect gaps, because Google's direct checkout fires only server-to-server pixels. ### When is this not worth your time yet? If you don't sell to the US or any of the countries above, most of this does not reach your shoppers today. And if your catalogue is mostly services, event tickets or other excluded categories, those products will not be in Shopify Catalog at all. None of this guarantees an AI channel will recommend you. Shopify's listing quality meter describes what generally helps a product rank, not a promise. What it does mean is that a store with complete product data and policies is one these agents can read properly. For app builders: Shopify's Global Catalog REST API stopped serving traffic on 2 November 2026, and apps that still call it need to move to Global Catalog MCP (Shopify developer changelog (https://shopify.dev/changelog/global-catalog-rest-api-stops-serving-traffic-on-november-2-2026)). If you'd like a hand checking your store's Agentic page, book a free 20-minute call (/book). ### We asked ChatGPT, Claude and Grok to fix three Shopify problems https://www.axioprax.com/insights/we-asked-chatgpt-claude-grok-shopify-questions (2026-11-05) We gave ChatGPT, Claude and Grok the same three Shopify problems and checked every step against Shopify's help pages. What they got right, what they got wrong, and what all three missed. Plenty of store owners now ask an AI assistant before they open the Shopify Help Center. So we gave ChatGPT, Claude and Grok the same three everyday Shopify problems and checked every step they gave against Shopify's own help pages. All three would have got a store owner most of the way there. None of them was perfect, the mistakes were small but real, and all three missed a Shopify change from the week before. We build automations with Claude, and Claude helped us draft this article. That is a reason to read the Claude results with care, so we scored every answer against Shopify's pages, not against each other, and we list each AI's mistakes in full below. ### How did we test them? We wrote three questions the way a store owner would, with the details that matter. These are the exact prompts: - Low-stock email. "I run a Shopify store on the Grow plan. I want to get an email when any product variant drops below 5 units in stock, without paying for an app. Give me exact step-by-step instructions in the Shopify admin, using the real menu and template names." - A card-testing attack. "Last night my Shopify store got about 40 small orders in an hour, most with failed card payments and a few that went through. I use Shopify Payments. What exactly should I do in the Shopify admin today? Give step-by-step instructions with the real menu paths." - A short payout. "My Shopify Payments payout this week was less than my sales. How do I match each Shopify payout to the deposit in my bank account, step by step, in the Shopify admin?" Each question went into a fresh chat with no history or memory, on 2 November 2026, using each product's default settings: Grok's default Fast model in a private chat, ChatGPT's default model in a temporary chat with medium thinking effort, and Claude's default Opus 5.5 model at medium effort in an incognito chat. All three cited web sources, and Grok and Claude showed the searches they ran. Claude asked permission once to read a Shopify help page, and we allowed it. We ran each question once, so treat this as a snapshot, not a ranking. Ask again tomorrow and the wording will change. To score, we opened the Shopify help pages behind each answer and checked every menu path, setting name and claim. We sorted problems into three kinds: wrong (Shopify's page says otherwise), not in Shopify's pages (we could not confirm it), and missing (a step that matters was left out). Our own guides for the same three problems are linked at the end of each section. ### Question 1: can they set up a low-stock email? Yes, all three. Each one sent us to Shopify Flow, which is free, and to the right template by its exact name, Get notified by email when product variant inventory is low (Shopify Help Center, Product variant inventory quantity changed (https://help.shopify.com/en/manual/shopify-flow/reference/triggers/product-variant-inventory-quantity-changed)). All three also added the step most quick guides skip: a second condition on the previous stock level, so you get one email when stock crosses the line instead of one after every sale. Shopify warns that if you check only the current amount, your reorder email is sent each time the product is ordered until you restock (Shopify Help Center, examples of workflows (https://help.shopify.com/manual/shopify-plus/flow/examples)). - ChatGPT: no errors found. It set the condition to less than 5 with a previous quantity of 5 or more, named the trigger correctly as Product variant inventory quantity changed, previously Inventory quantity changed, and added a table of which stock changes would send an email. - Claude: one error. The setup was right, but it said Shopify's docs call the trigger Inventory quantity changed and older guides use the longer name. It's the other way round: Shopify's page names the trigger Product variant inventory quantity changed and says it was previously called Inventory quantity changed. Harmless if you're following along in Flow, confusing if you're searching for it. - Grok: one slip. It set the condition to 5 or fewer, so the email arrives at 5 units, not below 5 as we asked. It did mention "less than" as an option. It was also the only one to point out that the trigger runs even when inventory tracking is off, which Shopify confirms. Our step-by-step version: get a low-stock email on Shopify without an app (/fixes/shopify-low-stock-alert-without-app). ### Question 2: do they know what to do after a card-testing attack? Mostly. All three recognised card testing straight away and gave the same core plan: check Shopify's fraud analysis on each order that went through, don't ship them, cancel and refund, turn on Decline charges that fail CVV verification under Settings, Payments, Shopify Payments, Manage, then switch to manual payment capture for a while. We checked those paths and setting names against Shopify's pages and they are right (Shopify Help Center, configuring Shopify Payments (https://help.shopify.com/en/manual/payments/shopify-payments/configuring-shopify-payments) and payment authorization and capture (https://help.shopify.com/en/manual/payments/payment-authorization)). - ChatGPT: the most complete answer, in eight steps. It named the Order risk evaluation section correctly, included the optional Send a warning 1 day before an authorization expires setting, the Blocked view in abandoned checkouts, Fraud Control rules and the chargeback filter on the Orders page. One step we could not confirm: the exact route it gave for reaching a human in Shopify's Help Center. - Claude: the best detail on cancelling. It warned that cancelling in bulk can refund orders that were already refunded, which Shopify's cancelling page states, and that card fees aren't returned on refunds. It also gave the 7-day authorization window for Shopify Payments, which is right. It left out the Blocked view in abandoned checkouts. - Grok: a good plan with the Blocked view included, but two claims we could not find in Shopify's pages: that reporting the attack helps your account standing, and its advice to unpublish or reprice the cheap product being targeted. The second may be sensible, it just isn't Shopify's guidance. What all three missed: on 28 October 2026, five days before our test, Shopify said Shopify Protect is on for every business on Shopify Payments, declines payments it is near-certain are fraud before they reach the bank, and gives each order a plain-language explanation of its risk (Shopify blog, Introducing Shopify Protect (https://www.shopify.com/blog/introducing-shopify-protect)). None of the three named Shopify Protect or its new risk explanations, though each said in general terms that Shopify already blocks some attempts. Their advice still works. It just doesn't tell you what changed that week. None of them named the free Flow template Capture payment if order is not high fraud risk either, which pairs with manual capture so you only check the risky orders by hand. Our step-by-step version: how to catch card-testing orders on Shopify (/fixes/catch-card-testing-orders-shopify). ### Question 3: can they match a payout to a bank deposit? Yes, and this was the strongest round. All three gave the right route, Finance, then View payouts in the Payouts section, said to open each payout and use its breakdown of charges, refunds, fees, adjustments and reserves, and said to export the payout transactions as a CSV. All three said deposits typically take 1 to 3 business days to arrive, which matches Shopify's page (Shopify Help Center, payout details (https://help.shopify.com/en/manual/payments/shopify-payments/payouts/view-details)). - Claude: the most precise, because it read Shopify's payout page before answering. It covered the older Money page for admins that still have it, the bank reference number to give your bank, retried payouts, and the Pending section for sales not yet paid out. No errors found. - ChatGPT: no errors found. It filtered by payout date, checked for Deposited status and Pending funds, and pointed to the Shopify Payments activity report if weekly totals still differ. - Grok: correct, and the only one to mention the Bank Reference column in payout exports, which Shopify's changelog shows was added in January 2026 (Shopify changelog (https://changelog.shopify.com/posts/better-way-to-match-shopify-payments-payouts-with-bank-deposits)). It called the bank reference a "Trace ID", which isn't Shopify's term, guessed that a bank might take a small fee, and didn't mention Pending, which is often why a week's payouts look short. Our step-by-step version: match Shopify payouts to bank deposits (/fixes/match-shopify-payouts-to-bank-deposits). ### So which one should a store owner use? On three questions and one run each, we would not crown a winner. ChatGPT had the fewest errors, Claude was the most exact when it read the source page, and Grok was quick and mostly right with a couple of loose claims. Their mistakes were small. What they all missed was a Shopify change from the week before. So use the steps they give you, then check them before you change anything in your store. - Open the source. All three cited Shopify pages. Click one and find the step there. That alone catches a reversed name or an off-by-one condition. - Match the exact names. Menus, templates and settings in your admin should match the answer word for word. If they don't, search the Help Center for the name you can see. - Test small first. Lower one product's stock, or cancel one order, before you change a setting for the whole store. - Check what changed recently. Shopify's changelog and its announcements cover the last few weeks, which is exactly where AI answers are weakest. ### What are the limits of this test? It was small on purpose: three questions, three assistants, one run each, on one day. Answers change between runs and between model versions, and paid tiers or other settings could do better or worse. We only scored claims we could check against Shopify's own pages. Anything we could not confirm is marked as such rather than called wrong. ### Is this order real? Fraud, bots and AI-agent orders on Shopify https://www.axioprax.com/insights/is-this-shopify-order-real-fraud-bots-ai-agents (2026-11-08) How to tell a fraud order, a bot and an AI-agent order apart in your Shopify admin, using Order risk, attribution and patterns, and what to do with each one. Three kinds of order make store owners stop and stare at the screen. A burst of tiny orders at 3 am. A customer list full of names nobody recognises. And, newer, an order that came in through an AI chat instead of your own website. The first is usually fraud, the second is usually bots, and the third is usually a real customer. Treating them the same way costs you money in both directions: you ship to fraudsters, or you cancel a genuine buyer. This piece shows how to tell them apart with what Shopify already gives you in the admin, and what to do with each. Every Shopify fact comes from Shopify's own help pages and announcements, opened on 2 November 2026. Where we mention our own tests, they ran on our demo store with a fictional shop and synthetic data. ### How do I check whether a single order is fraud? Open the order and look at the Order risk section. Shopify runs fraud analysis on online credit card orders that Shopify can verify and rates the risk of a chargeback due to fraud as low, medium or high. Medium and high risk orders are also flagged with a warning symbol on the Orders page (Shopify Help Center, fraud analysis (https://help.shopify.com/en/manual/fulfillment/managing-orders/protecting-orders/fraud-analysis)). The section also lists indicators, such as whether the card passed Address Verification System (AVS) checks, whether the customer entered the correct CVV, a location mismatch between the customer and the payment method, unusual device or network activity, and more than one card attempted. Shopify warns that individual indicators don't represent the order's overall risk level, so read the recommendation, not one red line. Two limits to know. The recommendation itself needs the Grow plan or higher, or Shopify Payments; indicators show from the Basic plan up. And some orders get no recommendation at all, including test, free, B2B, Point of Sale and subscription renewal orders. On 28 October 2026 Shopify widened Shopify Protect. It says Protect is on for every business on Shopify Payments, declines payments it is near-certain are fraud before they reach the bank, and gives every order that comes through a plain-language explanation of its risk (Shopify blog, Introducing Shopify Protect (https://www.shopify.com/blog/introducing-shopify-protect)). That means the orders you now see have already passed a stricter filter, and the ones flagged medium or high deserve a real look. ### What else gives a fraud order away? Shopify's own fraud prevention page lists signs worth checking on any order that feels wrong (Shopify Help Center, preventing fraud (https://help.shopify.com/en/manual/payments/fraud-prevention/preventing-fraud)): - The customer's IP address is far from where they say they are, belongs to a web hosting company, or is a proxy. - The phone number is invalid. Calling the customer is a fair test. - The email address appears in documented fraud reports. - The shipping address is far from the billing address. Gifts can legitimately differ, so this is a reason to look, not to cancel. - Several orders with different billing names and states ship to the same destination. ### What does card testing look like? Card testing is someone running stolen card numbers through your checkout to find which ones work. Shopify describes the pattern as many small payment attempts at once, attempts to save card details rather than buy, and a surge of declined transactions that can also push up your decline rate for real customers. Most attempts fail, so they never become orders. Shopify says checkout attempts it identifies as suspected card testing or bot activity aren't included in abandoned checkouts. The few that succeed are the problem: they look like small, normal orders, and if you ship them, the chargebacks follow. If you see a burst, our fix walks through it step by step: how to catch card-testing orders on Shopify (/fixes/catch-card-testing-orders-shopify). In short, switch card payments to manual capture, let a free Shopify Flow template capture only orders that are not high risk, decline charges that fail CVV, and block repeat sources with Fraud Control. ### Are bots placing orders in my store? Mostly not. Bots usually leave traces before an order: sessions in your analytics, junk customer accounts from sign-up forms, and checkouts nobody finishes. Shopify's bot filtering is about analytics. It applies only to session metrics and filters identified bot sessions out of session reports by default, and Shopify says it does not block bots from your site, because that would hurt search indexing and social sharing (Shopify Help Center, bot filtering (https://help.shopify.com/en/manual/intro-to-shopify/bots/bot-filtering)). So the bot problem shows up in your customer list and your email bill, not your orders. On our demo store, a Shopify Flow workflow tagged 17 of 17 imported bot-like customers as suspected-bot and none of the 4 genuine ones, and removed the bot tags when a tagged customer placed an order (our Bot Cleanup demo build (/insights/bot-cleanup-demo-build)). The rule we follow: tag, review, never delete on suspicion alone. ### What does an AI-agent order look like? Like a normal order with a different source. Shopify's agentic storefronts let shoppers find and buy your products in AI channels, and the feature is active by default for eligible stores. Google AI Mode and Gemini, Microsoft Copilot and Meta can take the order inside the chat through a Shopify-powered direct checkout when direct checkout is on, while ChatGPT sends the shopper to your own checkout. Shopify says orders from AI channels show in your admin with channel or referrer attribution, and orders customers place through personal AI agents show with attribution to the agent (Shopify Help Center, agentic storefronts (https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts)). Two things follow. First, an AI channel name on an order is not a fraud signal by itself. Behind it is a person who asked an assistant to buy something, and Shopify says you keep full ownership of that customer relationship. Second, the order still has a payment, so judge it the way you judge any order: its risk level, its indicators and its pattern. Shopify's agentic pages don't describe a separate fraud check for agent orders, so we would not assume one. Our full guide covers which channels sell, where the settings sit and how to switch them off: can AI agents buy from your Shopify store? (/insights/can-ai-agents-buy-from-your-shopify-store) ### How do I sort an order in under a minute? - Where did it come from? Check the order's channel or referrer attribution. Your online store, an AI channel and Shop are all normal sources. - What is the risk level? In the Order risk section, low risk with no warning symbol means fulfil as usual. - What do the indicators say? For medium or high risk, read the CVV and AVS results, location mismatch and card attempts, then the signs above. - Is it part of a pattern? Many small orders in a short time, or several names shipping to one address, point to card testing or a repeat fraudster. - Still unsure? Hold it. Shopify's Hold fulfillment order action in Flow changes the order's status to On hold, and orders with that status aren't fulfilled until the hold is removed (Shopify Help Center, Hold fulfillment order (https://help.shopify.com/en/manual/shopify-flow/reference/actions/hold-fulfillment)). Contact the customer, then release or cancel. ### What should I do with each kind? - A real order from any channel, including AI: fulfil it. Answer its questions like any other customer's. - A likely fraud order: don't ship it. For high-risk orders Shopify says you can verify the order, cancel it or refund it. Shopify also notes the bank that issued the card decides any reversal, not Shopify, so the earlier you stop a bad order, the better. - Card testing: follow the card-testing fix above, and check our repeat fraudster fix (/fixes/block-repeat-fraudsters-shopify) if the same details come back. - Bot sign-ups: tag and review, then suppress them in your email tool so you stop paying to email them. - Not sure: hold fulfilment for a named person to check. That is what our Fraud Hold kit (/store-kits/fraud-hold) sets up in Shopify Flow. ### When is this not enough? None of this guarantees you won't get a chargeback. Shopify says the decision to reverse funds is made by the bank that issued the card, and that Shopify doesn't cover charge reversals from banks. Shopify Protect is the exception, for the orders it covers. If you are already fighting a dispute, open the order in your admin, check the due date on the chargeback banner and send your evidence before it. And if you use a payment provider other than Shopify Payments, several of the tools here behave differently: Shopify says card testing protection and proxy detection are only for Shopify Payments stores, its 28 October announcement turns Shopify Protect on for businesses on Shopify Payments, and third-party gateways may not send fraud review updates back to Shopify. ### How one person runs e-commerce operations with AI https://www.axioprax.com/insights/one-person-ecommerce-operations-with-ai (2026-11-17) Three AI tools, one spreadsheet, one automation server and one Telegram chat. What runs on a clock, what still needs a person, what broke, and what it costs. ### Why I set it up this way Axioprax is me. The work I sell is Shopify operations, CRM and automation for stores in the US and UK. Since 28 October I have published a new fix page every day, written cold emails backed by a real ChatGPT test, answered questions on Reddit, posted on X, Threads, LinkedIn and YouTube, and checked all of it. No one person has the hours for that by hand. So I stopped asking "which AI tool should I use?" and started asking a duller question: which jobs repeat every day, and which ones need my judgment? Everything that repeats got a schedule, a place to write its output and a person (me) who sees the result. ### Who does what? - Claude runs two scheduled jobs a day, at 06:47 and 18:53. It writes the fix pages, drafts the social posts, makes the short videos, checks that every other job actually ran, and sends me a report. - Grok handles the lanes that live on X and in browsers: a trend brief each morning, posting on X and Threads, answering on Reddit in three shifts, finding people on LinkedIn worth talking to, and spotting companies whose job ads describe a problem we fix. - n8n, on a free Oracle Cloud server, runs the pipes: researching cold email leads, watching community forums for new questions, publishing the day's fix page at 12:15, and turning everything that needs me into a Telegram card with buttons. - Google Apps Script moves approved content between folders and logs anything that missed its slot at 22:30. - One Google Sheet is the shared memory. Every tool reads from it and writes to it. One tab, Lane Status, has a row for each of about 30 jobs, with when it last produced something and whether it is red. - Telegram is the only screen I need. Approvals, questions and reports all arrive there. ### What does a normal day look like? - 06:30 to 07:30. Grok's first Reddit shift and the trend brief. Claude's morning run checks the night's logs, fixes what it can and sends me a numbered list of what needs me. - 08:30 on weekdays. The cold email feeder drafts the day's emails. Each one includes a real test: a shopper's question about the store's own products, asked in ChatGPT, and the stores it named instead. Each draft arrives as a card. Nothing is sent until I tap Approve. - 12:15. The day's fix page goes live, unless I tapped Stop on the 11:00 card. - 16:40. That evening's posts arrive for review. At 17:45 they move to production unless I stop them. - 18:15 onwards. Grok posts on X and Threads. Claude's evening run checks what went out, what didn't, and why. - 22:30. Anything that missed its slot gets logged, so it shows up in the next morning's email instead of quietly vanishing. ### What do I still do by hand? More than people expect, and that is on purpose. - I approve every cold email, one tap per email. - I decide anything with a judgment call in it: whether to reply to a price question, whether a risky forum answer goes out, whether a catch-up post still makes sense a week late. - I record the screen recordings myself. The AI writes the script and builds the motion videos, but it cannot click through a real store for me. - I write and send LinkedIn messages myself. Agents only send a handful of connection requests a day, inside strict limits, because my profile is not something I let an agent put at risk. Most days that is a list of taps and a handful of real decisions. The rest of my day goes on work that needs a person. ### What broke in the first three weeks? Quite a lot. These are the ones worth knowing about if you try this yourself. - A forum warned us for undisclosed AI. The HubSpot Community marked one of our answers as spam because it was written with AI and didn't say so. It was the second warning, so I stopped answering there completely. Our Shopify Community answers now say they were drafted with AI help. - Reddit blocked the server's browser. Requests from a data-centre address hit rate limits, and answers that were ready never went up. They now get held and posted in the next shift instead of being lost. - Reports silently stopped reaching my phone. For a couple of days, any report containing an underscore or an angle bracket was rejected by Telegram's formatting rules. Nothing looked broken from the outside. I only noticed because a morning report didn't arrive. - The morning email cried wolf. It listed every task from the day before as "missed", including the ones that had run fine. A one-line filter fixed it, but for a few days it trained me to ignore the email, which is worse than having no email. - An agent undid a Like. While opening a list of people who reacted to a post, the browser clicked the wrong thing and removed my own Like. Harmless, but it is why agents now open those lists only from the post page. - A video went out looking unfinished. Text overlapped and half the screen was empty. I caught it, the video was rebuilt the same morning, and the first version never went public. ### Which rules keep it from going wrong? - Nothing public goes out without a human yes, or a human chance to say no. Cold emails need a tap. Posts and fix pages go out on a timer that I can stop. - An independent agent fact-checks every draft. It never sees the drafting. It opens every source and marks each claim correct, wrong or unsupported. On the first seven fix pages it found 7 wrong and 16 unsupported claims, all fixed before they went live. That one check is the reason I trust the rest. - Every job writes a log, and something reads it. If a job ran but wrote nothing, it is red on the Lane Status tab and shows up in the morning report. - One place to make decisions. Every question from every tool comes to the same Telegram chat as a card with buttons. If I have to go looking for a question, it doesn't get answered. - Say when AI wrote it. In forums and on this site. People don't mind help from AI. They mind finding out later. ### What does it cost to run? Far less than a part-time hire. Here is what a setup like this costs at current list prices (US, November 2026): - Claude Pro: $20 a month - ChatGPT Plus: $20 a month - SuperGrok: $30 a month - n8n Community Edition, self-hosted: free - A small server on Oracle Cloud's Always Free tier: free - Google Sheets, Apps Script and Telegram: free That comes to roughly $70 a month. Heavy daily use can push the Claude plan up to a Max tier, which starts at $100, so a realistic budget is $70 to $200 a month. The bigger cost is the setup time, and the discipline to read the morning report every day even when it is boring. ### Could a store owner run their store this way? Yes, and you can start smaller than this. Pick the three jobs you repeat every day: checking for stuck orders, answering the same five customer questions, and spotting stock that is about to run out are common ones. Give each one a fixed time, a place where its output lands, and one message to your phone when something needs you. Add the next job only when the first three have run for a week without you babysitting them. The free fixes on axioprax.com/fixes (/fixes) are a good place to find those first jobs. Each one is a single problem, with the settings and steps to fix it. ### Want this for your store? If you'd like help setting up the first three jobs for your store, book a free 20-minute call (/book). Bring the job you repeat most. This article was drafted with AI help and checked by me. ### How do I protect my Shopify store from card testing, chargebacks and payout holds before Black Friday? https://www.axioprax.com/insights/protect-shopify-store-card-testing-chargebacks-payout-holds (2026-11-18) Before Black Friday, set up Shopify's free tools against card testing, prepare for chargeback deadlines and check for payout holds or reserves, step by step. ### Why do card testing, chargebacks and payout holds show up together? Because each one feeds the next. A card-testing attack can leave you with real orders paid by stolen cards. Those cardholders can later file chargebacks. And chargebacks are one of the signals Shopify Payments watches when it decides whether to hold payouts or set a reserve. Shopify is direct about the first link. In its own words, card testing is when fraudsters run many small payment attempts at once to identify which cards are active. Cardholders whose details were used can later file chargebacks with their bank for any unauthorized charges. Each chargeback counts toward your chargeback rate regardless of whether you win or lose the dispute (Shopify Help Center, preventing fraud (https://help.shopify.com/en/manual/payments/fraud-prevention/preventing-fraud)). The second link is on Shopify's reserves page. The factors it lists for a reserve include more chargebacks, a higher refund rate and significant volume surges (Shopify Help Center, reserves (https://help.shopify.com/en/manual/payments/shopify-payments/payouts/reserves)). Its risk evaluation page also names rapid increases in sales volume. A big sale is exactly that, so tidy up before it starts. ### How do I stop card-testing orders from turning into chargebacks? Hold risky orders before money moves, and let Shopify's fraud analysis decide which ones wait for you. If you use Shopify Payments, Shopify lists card testing protection as built in. The steps below deal with whatever still gets through. - Check you are on Shopify Payments. Card testing protection and proxy detection are listed for stores that use it, and Shopify runs enhanced fraud analysis on its transactions. - Go to Settings, then Payments. In the Payment capture method section, select Manually and click Save. Shopify Payments gives you 7 days to capture an authorized card payment, so do not let orders sit (Shopify Help Center, authorization and capture (https://help.shopify.com/en/manual/payments/payment-authorization)). - Install Shopify Flow. Go to Apps, Flow, Create workflow, Browse templates, then Risk, and install Capture payment if order is not high fraud risk. Shopify says capture workflows do not work if automatic capture is still on (Shopify Help Center, high-risk orders with Flow (https://help.shopify.com/en/manual/orders/flow)). - Keep the Order risk analyzed trigger. Shopify warns against Order created for risk workflows, because fraud analysis takes time to finish after an order is placed. - In the Otherwise branch, add an Add order tags action or a Send internal email action, so high-risk orders that were not captured reach a person. - Go to Settings, Payments, Shopify Payments, Manage. In Fraud prevention, click Reduce credit and debit card fraud. Turn off Use automated settings, turn on Decline charges that fail CVV verification and click Save. Shopify says AVS checks can raise failed transactions, so think twice before turning that one on (Shopify Help Center, configuring Shopify Payments (https://help.shopify.com/en/manual/payments/shopify-payments/configuring-shopify-payments)). If an attack is already running, cancel the high-risk orders before you fulfil anything. Shopify notes that cancelling doesn't prevent the cardholder's bank from filing one, because the authorization hold can still show on the real cardholder's statement. Our card-testing fix (/fixes/catch-card-testing-orders-shopify) has the full walk-through. ### How do I block someone I already know is a fraudster? Use Shopify's free Fraud Control app. Its checkout rules stop a matching checkout before it becomes an order. Shopify offers it on all plans in places where Shopify Payments is available (Shopify Help Center, Fraud Control (https://help.shopify.com/en/manual/payments/fraud-prevention/fraud-control-app)). - Install Fraud Control from the Shopify App Store. - Go to Apps, Fraud Control, Rules, then Create rule. - Add the email, address or IP address under Checkout conditions. A rule can combine conditions. Shopify's own example pairs an IP address with a ZIP code, which narrows what the rule blocks, so legitimate checkouts are less likely to be caught. - Click Save. The rule is active straight away. Shopify names it from its conditions, and you cannot rename it. - From the Fraud Control app page, look at the Flow templates Shopify offers. Two are useful for this week: Notify your team of a chargeback and Restrict orders from customers to five a day. The buyer is not told they matched a rule. They see an error that checkout could not be completed and are asked to contact the store. Blocked checkouts show in abandoned checkouts. Fraud Control also shows your chargeback health if you use Shopify Payments. Shopify calls chargebacks under 0.4% of orders good standing, 0.4% to 0.6% at risk, and over 0.6% its highest risk level, where your risk is higher than average. Step by step: our repeat fraudster fix (/fixes/block-repeat-fraudsters-shopify). ### What should I do the moment a chargeback arrives? Find it, note the due date and gather evidence that matches the reason. Do not try to refund. Shopify says you cannot refund a chargeback once it has started. The disputed amount and a chargeback fee leave your account straight away (Shopify Help Center, responding to chargebacks (https://help.shopify.com/en/manual/payments/chargebacks/chargeback-process)). - Go to Orders, click Search and filter, add the Chargeback and inquiry status filter and select Open. - Open the order and read the banner. Write down the reason and the due date. Shopify shows it in your store's timezone, and if no time is shown, the deadline is 11:59 PM on that date. The due date is 7 to 21 days after the chargeback is filed (Shopify Help Center, chargebacks in the admin (https://help.shopify.com/en/manual/payments/chargebacks/chargebacks-in-admin)). - Look for a Not enough evidence to challenge this chargeback notice. If you see it and do nothing before the due date, Shopify accepts the chargeback automatically. - Gather proof for that reason. For product not received, Shopify lists tracking, delivery confirmation, proof it went to the checkout address and carrier notifications. Each file must be a PDF, JPEG or PNG under 2 MB, with 4 MB in total. - Click Add evidence in the banner, put your strongest proof first and click Save. Only click Submit now when everything is in, because you cannot edit afterwards. Shopify adds some order details automatically and submits on the due date even if you add nothing. Do not rely on that alone. Shopify's own line is that the bank and card company decide the outcome, not Shopify. Strong evidence helps, but Shopify says it doesn't guarantee that you win. Full guide: our chargeback fix (/fixes/respond-to-chargeback-shopify). One setup task belongs before the sale. Shopify has an Order dispute updates staff notification that controls who is emailed about disputes. Check it points at an inbox someone actually reads during sale week. ### How do I stop the same customer causing a second chargeback? Tag them when the dispute opens, and hold their next order. Shopify Flow has a Dispute created trigger that runs when a chargeback or inquiry is created on an order that uses Shopify Payments. Shopify lists Add customer tags and Hold fulfillment order among the actions it works with (Shopify Help Center, Dispute created (https://help.shopify.com/en/manual/shopify-flow/reference/triggers/dispute-created)). - In Apps, Flow, create a workflow with the Dispute created trigger. - Add an Add customer tags action with a neutral tag such as dispute-review. - Build a second workflow of your own (there is no Shopify template for this) that checks new orders for that customer tag and uses the Hold fulfillment order action, so a person checks them before they ship. - Tag customers from disputes that opened before you turned this on, because the trigger only runs for new ones. ### Why would Shopify hold my payout or set a reserve, and what can I do? A hold usually stops the payout while Shopify checks something. A reserve keeps back part, or sometimes all, of your money for a set period to cover chargebacks and refunds. You can prepare for either one. You cannot decide either one. Shopify does. Shopify's reasons for account holds include missing information, wrong bank details, product eligibility and business types it does not support (Shopify Help Center, account holds (https://help.shopify.com/en/manual/payments/shopify-payments/payouts/account-holds)). On the reserve side, Shopify uses two types. A fixed amount reserve holds a set sum for a defined period. A percentage-based reserve holds a share of each transaction. Shopify's example is 10% for 120 days, with the other 90% paid out as normal. Before the sale: - Go to Settings, Payments and look at the Shopify Payments status indicators. Green means working as expected. Yellow means that part is not working as expected, or is expected to stop working. Red means that part is blocked. - Go to Finance, then Payouts. Pending is captured money not yet added to a scheduled payout. Reserve appears only if your account has reserved funds. - Confirm which email is the store owner's in Settings, Users and permissions. Shopify sends hold and reserve notices there and to Merchant Alerts in your admin. Only the store owner can see the account details and resolve a hold. - Check that inbox and its spam folder now, not after a payout fails to arrive. - Leave your bank account alone if you can. Shopify says changing to a new external bank account pauses payouts for 3 to 5 business days. If a notice does arrive, Shopify says the fastest route is to send what it asks for and reply directly to its email. Customers can usually still check out during a hold, but you are not paid for those orders until it is resolved. A reserve has an expiry date in its email. When it expires without an extension, Shopify says the money is included in upcoming payouts and might take several business days to process (Shopify Help Center, risk evaluation (https://help.shopify.com/en/manual/payments/shopify-payments/payouts/risk-evaluation-process)). Our payout hold fix (/fixes/shopify-payout-on-hold-what-to-prepare) covers this in detail. ### How do I test all of this before real customers arrive? Use Shopify Payments test mode, then switch it off. Shopify says you should not turn on test mode on a store that is in production: no real payments are captured while it is on, and customers paying with Apple Pay or Google Pay can still complete checkout without being charged (Shopify Help Center, testing Shopify Payments (https://help.shopify.com/en/manual/payments/shopify-payments/testing-shopify-payments)). If you only have your live store, keep the window as short as you can, at your quietest hour, and look for orders with the test mode banner afterwards. - Go to Settings, Payments, Shopify Payments, Manage. In Test mode, select Enable test mode and click Save. - Place an order with Shopify's test Visa, 4242424242424242, a name of at least two words, any future expiry and any three digits. - Check that your capture workflow ran and the order was captured or tagged as you expect. - Place a second order with 4000000000000259, Shopify's test card for a disputed transaction. Shopify names test mode as a way to test the Dispute created trigger, so check your dispute workflow tagged the customer. - Turn test mode off and click Save straight away. Full checklist: our test checkout fix (/fixes/test-shopify-checkout-without-real-money). ### What can none of this do? We want to be plain here. None of these steps lifts a payout hold, ends a reserve early or wins a chargeback. Shopify decides holds and reserves. The card company decides chargebacks. Fraud Control rules only stop checkouts that match what you typed in. Shopify says the rules don't guarantee that chargebacks won't happen. A determined fraudster can change email, address and network. Most of this also depends on Shopify Payments. The Dispute created trigger and card testing protection need it, and Fraud Control is offered only where Shopify Payments is available and shows chargeback data only to stores that use it. Shopify Protect, which covers eligible orders against fraudulent and unrecognized chargebacks, is listed for US stores that use Shop Pay. If you use another payment provider, check what fraud and dispute tools that provider offers. And a note on our own pages. Our fix pages are written from work on a demo store with synthetic data. They show the settings work as Shopify describes. They do not show results on a live store. ### Where does the Protect Pack fit? If you would rather not set this up yourself, our Protect Pack (/store-kits/protect-pack) does it for you. It installs our Fraud Hold and Bot Cleanup kits together, plus card-testing safeguards and a check of your stock, checkout and payout settings before your busy weeks. It costs $499, one-off, which is $199 less than buying the two kits separately. It comes with a 30-day fix-it guarantee. Not sure what your store needs? Book a free 20-minute call (/book). This article was drafted with AI help and checked by me. ### Stocky Replacement: Shopify Reordering on a Demo Store https://www.axioprax.com/insights/stocky-rescue-demo-build (2026-10-20) How we replaced the Stocky reorder routine on our Shopify demo store with a supplier sheet, a reorder sheet, a Monday email and a native purchase order. Shopify says Stocky is no longer available as of 31 August 2026, and stores have at least 90 days after that to export their data. Suppliers can't be exported and old purchase orders can't be imported, so the reorder routine has to be rebuilt. This is how we rebuilt it, with a supplier sheet, a reorder sheet, a Monday email, an invoices tab, a stocktake tab and Shopify's own purchase orders, on our own demo store. It is for merchants who used Stocky for purchase orders, suppliers or stocktakes. Our kit, KIT-01 Stocky Replacement, went through several rounds of testing by a separate tester who had not seen the build, and the rounds it failed are listed below. What the kit includes and its price are on the KIT-01 Stocky Replacement page (/store-kits/stocky-rescue). If your Stocky read-only screens still open, our free checklist for old Stocky data (/store-kits/stocky-rescue#export-checklist) lists what to export, screenshot and write down first. It is a one-page PDF and needs no email. This is a demo build, not a client project. Everything below was built and tested on Northfield Outfitters, our own Shopify development store: demo store, fictional shop, synthetic data, real tools. There are no client results in this article. ### Key points - Shopify says Stocky is no longer available as of 31 August 2026, with read-only export access for at least 90 days after that date (Shopify Help Center, transitioning from Stocky, checked 20 October 2026). - Suppliers can't be exported from Stocky, and historical Stocky purchase orders can't be imported into Shopify (same page). - Stock on a Shopify purchase order is received through an inventory transfer created from the order once it is marked Ordered (Shopify Help Center, purchase orders). - On our demo store, with synthetic data, the Monday email's $7,550 supplier order matched Shopify purchase order #PO2 exactly. ### What happens to my Stocky data? Stocky is no longer available. If your store still has read-only access, you can export old data for now. Stocky is no longer available as of August 31, 2026. Merchants have read-only access to export your data for at least 90 days after that date (Shopify Help Center, transitioning from Stocky). The page gives no end date, so the window closes no earlier than about 29 November 2026, and it may stay open longer. The same page has two lines that shape most of the work. Suppliers can't be exported from Stocky. Historical purchase orders can't be imported into Shopify either. So the supplier list gets rebuilt by hand, and any old orders you exported stay in files you keep. On the Shopify Community, merchants have listed what went with it: purchase order export, stocktakes, marking invoices paid or unpaid, and invoice due dates (reported by merchants in Shopify Community thread 639203, read October 2026). ### Why can a Shopify app only see 60 days of orders? Standard app access (read_orders) reaches back only 60 days. Older orders need read_all_orders, which Shopify has to approve (shopify.dev, access scopes). The kit avoids this by using Shopify's own sales report export instead of an app, so the reorder sheet can work from 90 days of sales. On our demo store that report had no usable product history, so we used synthetic sales (see below). Update, 2 November 2026. Shopify's purchase orders now hold suppliers with their contact details and address, and Shopify has built-in stock counts that a manager approves (Shopify Help Center, purchase orders and stock counts, checked 2 November 2026). The kit now sets up your suppliers there and runs your first stock count in Shopify. The supplier sheet keeps only what Shopify does not, such as lead times and minimum orders, and the Stocktake tab below becomes an optional review step before a count is approved. ### What does the Stocky Replacement kit include? Seven parts that bring back the weekly reorder routine, including the invoice tracking and stocktakes that merchants said they lost, plus an optional archive of old Stocky data. The rest of Stocky's features are out of scope. - Supplier sheet in Google Sheets or Excel, with contacts, lead times, minimum orders, payment terms and products supplied, built from your own records, any Stocky exports you have, and a short interview with whoever ran reordering. - Reorder sheet. One row per product with daily sales, stock, lead time, safety stock, reorder point and suggested quantity, plus an ORDER NOW flag. Rows with doubtful data are flagged and kept out of the order. Each Monday someone copies On hand and Incoming from Shopify into it, unless the store takes the Auto stock sync add-on (see below). - Monday reorder email. Products at or below their reorder point, grouped by supplier and checked against each minimum, with a short section on supplier invoices that are overdue or due in the next 7 days. It is set up as a weekly Google Sheets (Apps Script) email that runs from the store's own Google account. It has been run by hand from a real Google account on the demo workbook (19 October 2026), and the email arrived. It then ran on its own weekly schedule on Monday 26 October 2026, and that email arrived too. We have not built a Shopify Flow version. - Invoices tab. One line per supplier invoice: PO number, invoice number, date, amount, due date, Paid or Unpaid, and an overdue flag. Leave the due date blank and the sheet works it out from the supplier's payment terms. Open invoices are entered with you from your own records; Stocky's invoice history stays in any Stocky export you saved rather than being typed in again. - Stocktake tab. One line per product with what the Stock tab expects, what you counted, the difference and its value at cost, marked MATCH, OVER, SHORT or NOT COUNTED. It never changes Shopify: you review the count, then adjust stock in Shopify yourself. - Purchase order checklist for raising, sending, receiving and closing a Shopify native purchase order, with screenshots from the store's own admin. - Handover. A recorded walkthrough and a one-page "how this works / what to do if" note. - Stocky archive, only if your export screens still open. Every Stocky report still on offer, saved to a dated folder the store owns, with screenshots of supplier records that can't be exported. Shopify's Help page names completed purchase order reports as CSV files, stocktake history and historical cost data; we also take any other report the store's Stocky screens still offer. If your Stocky export screens still open, do the archive first, because it can't be redone once they close. There is one optional extra. The Auto stock sync add-on, at $149, copies On hand and Incoming from Shopify into the Stock tab early every Monday, before the email runs, so nobody has to do it by hand. We have built it as a Google Apps Script that runs in your own sheet, with an n8n workflow as a second option for stores that already use n8n. It only reads Shopify, and if anything looks wrong it writes nothing and says why. It is still in testing: so far it has run only against a simulated Shopify and Google Sheets, not a real store, so we don't offer it yet. ### What did we build it on? Northfield Outfitters, a fictional shop on a Shopify Plus development store of ours. It has never had Stocky, so we made up the supplier and sales data a Stocky store would have had. Of its 26 products, 25 had no SKU, so the sheets match on product and variant name. Its 32 orders had come in through our own n8n sync as custom line items, and Shopify's units sold report filed all 36 units under a product called "None". Shopify also stores a supplier only as a vendor name. So the demo uses the store's 18 tracked variants and their real on-hand stock, read on 16 October 2026, with synthetic sales and details for 4 suppliers over 90 days. Two snowboards with stock sat in Draft and Archived, so we left them out, and the kit now takes active products only. We also planted five gaps on purpose: a product with no supplier, a supplier with no lead time, a supplier with no email, a duplicate sales line and a product with no sales line. ### Why is the order in this article #PO2? Because we threw away our first run. Our first sample data had one board selling 450 units in 90 days and costs at full retail price, so the Monday email asked for $122,035 across two suppliers. We had already raised one of those orders in Shopify as #PO1, for $54,535, and received all but two units. The formulas were right, but no small shop sells boards in those numbers, so a demo built on them would have proved very little. We rebuilt the sales at small-store levels (2 to 40 boards per product over 90 days, 72 for the best-selling wax), set costs at about 45% of retail and gave one supplier a $150 minimum. In Shopify we rejected the two outstanding units, reset on hand for the three boards to 10, 10 and 0, raised #PO2 and archived #PO1, which hides an order without deleting it. ### What did the reorder sheet catch? All five planted gaps. Four of them held 10 products out of the order, and the fifth, a supplier with no email, was listed separately. Eight products showed MISSING DATA, one CHECK DUPLICATE and one NO SALES DATA, which left only 2 at ORDER NOW. Once we filled the gaps, flags dropped to 0 and ORDER NOW rose to 4. Sales hadn't changed. The Out of Stock Snowboard joined the order once it had a supplier, and the ski wax once its supplier had a lead time. The missing data had been hiding two orders that were already due. ### What does the Monday email say? What to order from each supplier this week and whether each order clears its minimum, with data fixes listed separately. Before the fixes it showed one order, Snowboard Vendor at $3,150, with 10 products to fix and 1 supplier with no email. Afterwards Snowboard Vendor came to $7,550, which meets its minimum. Our second sample supplier, which we had named Northfield Outfitters after the store, came to $132 of wax. That is under its $150 minimum, so the email says to top up or wait. ### Has the email run from a real Google account? Yes, once, by hand. On 19 October 2026 we ran the Monday email from a real Google account on the demo workbook, and it arrived with the right figures: two supplier orders (Snowboard Vendor at $7,550, meeting its minimum, and Northfield Outfitters at $132, below it) and one overdue invoice. The first attempt failed, because the script was not attached to the sheet, and the failure alert email arrived, which is what it is for. That run was started by hand. On Monday 26 October 2026 it ran again on its own weekly schedule, with nobody starting it, and the email arrived with that week's figures: two supplier orders, two overdue invoices and a warning that the stock numbers were 10 days old. If you only need a simple low stock alert, our list of n8n e-commerce automation workflows (/insights/n8n-ecommerce-automation-workflows) covers that smaller job, and our guide to Shopify automation workflows (/insights/shopify-automation-workflows-guide) covers other routine store tasks. ### Did the purchase order match the email? Yes, to the cent. We raised #PO2 to Snowboard Vendor for 6 x Dawn at $315, 4 x Electric at $315 and 11 x Out of Stock Snowboard at $400, a total of $7,550.00, then marked it Ordered. ### How do you receive stock on a Shopify purchase order? You create an inventory transfer from the purchase order once it is marked Ordered, then receive against the transfer. You don't receive on the purchase order itself (Shopify Help Center, purchase orders). Our first checklist got this wrong, pointing to a Receive inventory button that appeared nowhere in our screenshots, and our tester failed it. The rewrite follows what we clicked on the demo store: Create transfer, Mark as in transit, then Receive items. ### What happens to a short delivery on a Shopify purchase order? Shopify keeps the missing units In transit on the same transfer and shows them as Incoming, so the shortfall stays visible. We received 6 of 6, 3 of 4 and 11 of 11, as if one Electric board was late. Electric on hand went from 10 to 13, and the fourth board stayed In transit until we received it, when on hand reached 14. Type what turned up on each line, and use Accept all only when every unit is there. If the rest will never come, reject those units so they stop counting as Incoming. ### When is a Shopify purchase order closed? When every unit has been received or rejected. Shopify then shows the order as Transferred, which is as closed as it gets. Once the late board was in, #PO2 read Ordered and Transferred. ### What caught us out in Shopify's purchase orders? Three behaviours we had not expected now have their own lines in the checklist. ### Untracked products are greyed out When we added products to our first order, some were greyed out and labelled "Inventory not tracked". During setup we switch tracking on for anything you reorder, before the first order goes in. ### The last cost comes back on its own This one nearly put the bad data back. When we opened #PO2, Shopify had pre-filled each line with the last cost used for that product, which was $700 from our first run's #PO1. We retyped every line. When prices move, the pre-fill carries the old one forward, so check each line against the supplier's current price before marking the order Ordered. ### Sending and paying are on you Shopify does not send the purchase order to your supplier, so you send it by their usual route. Its purchase orders don't track whether an invoice is paid either, which is one of the things merchants said they lost with Stocky. So the checklist has you add each invoice to the Invoices tab when it arrives and mark it Paid when you pay it, and the Monday email reminds you about anything overdue. ### How hard did we try to break it? We ran five rounds with a separate tester working from a fresh copy, who had not seen the build or its logs. The kit failed the first two. Round 1 found 3 blockers: typing "10 units" into an on-hand cell made that board vanish from the email without warning, a product entered twice in the Stock tab was ordered twice, and a formula range stopped at row 19, so a product added below it was never ordered. Round 2 found that the workbook covered 500 products while we offered the kit to stores with up to 1,500. Smaller finds included a name starting with "#" read as a spreadsheet error, and the question mark in "Wax?" acting as a wildcard. Across the five rounds the tester logged 35 distinct issues (3 blockers, 5 major, 27 minor; an issue found again in a later round is counted once), and we fixed all of them. Rounds 3, 4 and 5 passed. The last 3 fixes were checked only by our own 17-case regression run. When we added the Invoices and Stocktake tabs, a separate tester who had not seen the build ran a sixth round on them. It passed, with 13 minor issues: an invoice whose supplier was typed in lower case was overdue in the sheet but missing from the email's overdue list, a credit note showed as overdue, and a count of minus five was accepted as SHORT, among others. We fixed all 13. A seventh round retested those fixes, and all 13 held. It found 10 more minor issues in the kit, and it failed the round on our marketing material: an older set of posts and videos, written before the kit grew, still carried wording we no longer use. We marked those files superseded. An eighth round failed too: our internal delivery notes still pointed at the first version of the workbook, and there was no blank copy ready for a client, so a client could have been handed the wrong file. It also found smaller issues, such as a supplier name containing * or ? being read as a wildcard. We fixed all of them. A ninth round retested those fixes and passed, with minor notes: for example, the delivery notes did not say plainly that a store with two locations must add both locations' stock together, or that the client, not us, should set up the weekly email schedule from their own Google account. We fixed those notes too; the fixes from the ninth round have so far been checked only by our own regression run. All of this was in-house testing on the demo store with synthetic data. Nobody outside has audited it. On a real store, you sign off against eight acceptance tests, from checking the supplier sheet to raising and receiving one test purchase order with the checklist alone, plus one overdue test invoice and one stocktake line that is deliberately out. ### What the kit does not do - It isn't an ERP or a full inventory app. If you need landed costs, manufacturing or multi-warehouse transfers, an inventory app is the better fit, and we will tell you so. - Forecasting is a simple sales-rate reorder point, with no seasonality. - We don't set up a paid inventory app for you, and catalogue clean-up goes only as far as the sheets need. - On its own it does not read stock from Shopify. Skip the Monday copy and an order already placed is suggested again. The Auto stock sync add-on is built to do that copy, but it is still in testing. - Stock is one figure per product across up to 2 locations. Reorder points per location are custom work. - The Stocktake tab doesn't change stock in Shopify. You make the adjustment yourself after review. ### Is the kit a fit for your store? - You used Stocky for purchase orders, suppliers or stocktakes. - Stocky does not need to open. If its read-only export screens still do, we also save your old reports (see the FAQ below). - You have up to 2 locations, up to 1,500 active products and up to 40 suppliers. Bigger stores get a custom quote. - The products you reorder are active, with inventory tracking on or ready to switch on. We work through a Shopify staff or collaborator account with the right permissions, never an owner login, and no passwords by email. The kit is $349. If your store upgrades to Inventory Intelligence ($750) within 60 days, the $349 you paid for the kit counts towards it. ### Getting started If your store used Stocky and you want the Monday reorder routine back, with a Google Sheet, a Monday email and Shopify's own purchase orders, see what is in the KIT-01 Stocky Replacement kit (/store-kits/stocky-rescue), then describe your store at axioprax.com/start (/start?kit=stocky-rescue). Tell us roughly how many products, suppliers and locations you have, and whether Stocky still opens for you. We confirm eligibility before you pay. ### Frequently asked questions ### What replaces Stocky? For most small stores, Shopify's own purchase orders plus a reorder routine. Shopify has native purchase orders with saved suppliers, transfers and stock counts, but its help pages for them don't describe suggesting what to reorder. Our kit replaces the reorder routine with a supplier sheet, a reorder sheet and a Monday email. If you need landed costs, multiple warehouses or demand forecasting, a full inventory app is the better fit. ### Stocky won't open for my store any more. Can you still help? Yes. The kit does not depend on Stocky. We build the supplier and reorder sheets from your Shopify sales export and your own supplier records, at the same price, confirmed in writing before you pay. Only the optional archive of old Stocky reports needs Stocky to open. ### Can I import my old Stocky purchase orders into Shopify? No. Shopify does not support importing historical purchase orders, so they are kept as exported files in your own archive folder. ### Do my products need SKUs? No. The sheets match on product and variant name instead, as they did on our demo store. The handover note still recommends adding SKUs. ### Do I need Shopify Flow or a particular Shopify plan? You don't need Shopify Flow: the Monday email runs from Google Sheets, and we have not built a Flow version. Shopify's help page does not list which plans include native purchase orders, so we check your store before you pay. ### Will the sheet know about stock I have already ordered? Only if Incoming is copied in from Shopify each Monday along with the date, by hand or by the Auto stock sync add-on, which is still in testing. In our simulated runs, a stock date more than 8 days old put a STALE warning in the email. ### What this article is built on Every vendor claim was read again against its own page on 20 October 2026, and each community report is labelled as a report. Every demo figure comes from our build, capture and test logs for the demo store, fictional shop, synthetic data. Claim ledger. Demo figures are from a demo store, fictional shop, synthetic data, not a client. #ClaimSourceCheckedStatus 1Stocky no longer available as of 31 August 2026; read-only export for at least 90 days afterShopify Help Center, transitioning from Stocky20 Oct 2026Verified 2Suppliers can't be exported; historical purchase orders can't be importedSame page20 Oct 2026Verified 3Merchants report losing PO export, stocktakes, invoice paid or unpaid, and due datesShopify Community, thread 639203Oct 2026Reported by merchants 4"The hard part is getting your history out in one piece."Shopify Community, thread 651074Oct 2026Quoted from a community post 5read_orders covers the last 60 days; read_all_orders needs Shopify approvalshopify.dev, access scopes20 Oct 2026Verified 6Stock is received through a transfer created from an Ordered purchase orderShopify Help Center, purchase orders; demo store20 Oct 2026Verified 7Which Shopify plans include native purchase ordersHelp page does not list plans20 Oct 2026Unknown, checked per store 8Untracked products can't be added to a purchase orderDemo store16 Oct 2026Observed 9Monday email from a real Google accountRun by hand on the demo workbook, email delivered with the demo figures; failure alert also delivered19 Oct 2026Manual run observed; scheduled run observed 26 Oct 2026 10Stocky archive stepStocky can no longer be installed16 Oct 2026Not shown in the demo 11First sample data gave a $122,035 email and #PO1 at $54,535; rebuilt with costs at about 45% of retailDemo store build log, synthetic data16 Oct 2026Observed, run 1 superseded 1235 distinct issues over 5 test rounds (each counted once), all fixed; last 3 checked by a 17-case regression runIn-house test reports, demo store16 Oct 2026Observed 13Shopify does not send a purchase order to the supplier or track invoice paymentDemo store16 Oct 2026Observed 14Round 6 on the Invoices and Stocktake tabs passed with 13 minor issues, all fixed; round 7 retested them (all 13 held) and failed on old marketing material; round 8 failed on delivery notes pointing at the first workbook; round 9 retested those fixes and passed with minor notes; every round's issues fixedIn-house test reports, demo store20 Oct 2026Observed; round 9 fixes checked by our own regression run only 15Auto stock sync add-on reads Shopify and writes the Stock tabSimulated Shopify and Google Sheets only19 Oct 2026In testing, not yet run on a real store 16Stocky data Shopify says to export: completed purchase order reports as CSV files, stocktake history and historical cost dataShopify Help Center, transitioning from Stocky20 Oct 2026Verified ### Shopify Fraud Hold: Holding Risky Orders on a Demo Store https://www.axioprax.com/insights/fraud-hold-demo-build (2026-10-21) Shopify scores orders for fraud risk but nothing holds a risky one by default. How we held test orders with Shopify Flow until a person checked them. Shopify runs a fraud analysis on online card orders it can verify and gives each one a recommendation. What it doesn't do by default is stop a risky order from being packed. None of the risk templates Shopify offers in Flow holds fulfilment, so a store that ships from the unfulfilled list can send a flagged order out the same afternoon. Here is how we built a hold for that gap with Shopify Flow on our own demo store, what happened to the test orders, and what we couldn't show. The kit, KIT-04 Fraud Hold (/store-kits/fraud-hold), doesn't win disputes or make fraud go away. It puts a person between a risky order and the shipping label. Update, 2 November 2026. Shopify widened Shopify Protect on 28 October 2026: on Shopify Payments it now declines payments it is near-certain are fraud before they reach the bank, and explains each order's risk in plain language (Shopify blog, Introducing Shopify Protect). It also launched a separate app, Shopify Protect: Rule Builder, which can hold an order for review for $0.07 per screened transaction and opened with a waitlist (Shopify App Store listing, checked 2 November 2026). Orders Shopify is less sure about still come through, which is the gap this build covers with free Shopify Flow. This is a demo build, not a client project. Everything below was built and tested on 21 October 2026 on Northfield Outfitters, a fictional shop on a Shopify Plus development store of ours, with synthetic data and test orders we placed ourselves. There are no client results in this article. ### Key points - Shopify Flow's Order risk analyzed trigger starts a workflow when Shopify's fraud analysis of an order is complete, and Flow has a Hold fulfillment order action (Shopify Help Center). - On our demo store, one workflow held 3 of 4 test orders, tagged them fraud-hold and ran its email step to a reviewer address. The order that matched no signal went through untouched. A fifth order, placed after we reworded the hold note, was held too. - One order was held only because its billing country (Canada) differed from its shipping country (US). A person released it by hand and swapped the tag to fraud-reviewed. - A second workflow lists the orders still tagged fraud-hold each day. Its test run listed the three held orders. After the release, the same search returned two. ### The order that looks like a sale Shopify's fraud analysis gives an order a recommendation (accept, investigate or cancel) and a risk level, and Flow can read both (Shopify Help Center, Order risk analyzed; shopify.dev, OrderRiskSummary). Shopify also says card-testing attempts are mostly failed transactions, but some might result in placed orders (Shopify Help Center, preventing fraud). Those are the ones that matter here, because they look like sales. One merchant described the rest. In a Shopify Community thread posted on 27 August 2026, a UK merchant reported a fraud order that came after failed payment attempts and that Shopify had classed as low risk (thread 672811). Other merchants replied that a low rating is about chargeback risk, not a promise. One store's report proves nothing on its own, but it is why this kit also checks signals the store picks for itself. Then there is the paperwork. For a fraud dispute on Shopify Payments, the evidence that helps includes address and security code checks, proof of delivery to the verified address, 3D Secure, device and IP details, and messages with the customer (Shopify Help Center, chargeback process). The deadline is typically 7 to 21 days after the chargeback is filed, and in the US the fee is $15, returned if you win, varying by country. Holding a suspicious order gives the store a chance to cancel before anything ships. The bank still decides. ### What the kit includes The hold rule is the centre of it. Everything else, from the daily digest and review checklist to the dispute evidence template and handover, exists so that a held order gets looked at. The full list is on the kit page (/store-kits/fraud-hold). For other jobs Flow can take on, see our guide to Shopify automation workflows (/insights/shopify-automation-workflows-guide). ### The demo store and its limits Northfield Outfitters, the fictional shop from our Bot Cleanup kit, runs on a Shopify Plus development store with Shopify's test payment gateway, not Shopify Payments. That shaped the whole test. To give a test order a simulated risk level, Shopify says you need Shopify Payments in test mode (Shopify Help Center, fraud analysis), which meant entering business and card details for a demo store. We decided not to. Every order below went through the test gateway with card "1", made-up addresses in Denver and Toronto, and our own email addresses, and each one's risk panel reads "Analysis not available". So of the rule's four signals, Shopify's recommendation and risk level had nothing to read. The store's own two, a new customer above a spending threshold and a billing country that differs from the shipping country, could be tested properly. ### Building the hold in Shopify Flow It is one workflow, "Fraud hold: hold and notify", and it starts on Order risk analyzed rather than Order created. Shopify says that trigger runs once the fraud analysis is complete, and that third-party fraud apps don't fire it (Shopify Help Center, Order risk analyzed). The condition is a list joined by OR, so any one match is enough: - Shopify recommends cancel or investigate. - Any risk assessment on the order is high or medium. - The customer has one order or fewer and the total is $200 or more, grouped with AND. The $200 is a demo figure; on a client store the owner signs off the number. - The order has a shipping country, and the billing country is not equal to it. On a match, Flow adds the tag fraud-hold, loops over the order's fulfilment orders and holds each one with the reason Other and the note "Held for fraud review", then sends an internal email to the reviewer. That email action takes fixed addresses only (Shopify Help Center, Send internal email), so the address is written into the workflow. Sidekick, Shopify's assistant in Flow, built it from a written description, and we opened every step before turning it on. The logic was right. It didn't rename the workflow, though, which is why the checking step stays in the runbook. We expected trouble with the country check. There was none: on this Plus store, Flow compared the billing and shipping country fields directly. We haven't checked Basic or Grow. ### Hold, don't cancel A held order costs a delay. A wrongly cancelled one costs a customer. Flow offers a "Cancel and restock high risk orders" template, and we leave it alone. A hold stops fulfilment while the payment and order stay as they are, and Shopify keeps the inventory reserved (Shopify Help Center, fulfilment holds). ### Four test orders, one at a time We placed four orders a few minutes apart on the morning of 21 October, each from a new test customer. ### Order #1034: new customer, $749.95 One snowboard, both addresses in the US, so only the new-customer signal should match. It did. The order showed On hold, with a Flow hold reading "Other" and "Held for fraud review", and the fraud-hold tag. ### Order #1035: new customer, $32.95 The control. It matched nothing: no tag, no hold, no email. One clean order isn't a sample, so we won't call it a false-positive rate. ### Order #1036: billing Canada, shipping US Ski wax, $32.95 with shipping, from a new customer. Far below the $200 line, so only the country mismatch could hold it. It was held and tagged. A spending threshold alone would have let it through. ### Order #1037: two snowboards, $1,379.90 This was meant to be the split-location test. Both boards turned out to be stocked at one location, so Shopify created one fulfilment order and Flow held it. Shopify says each fulfilment order is held separately, and a partly unfulfilled order shows as Unfulfilled, not On hold (Shopify Help Center, fulfilment holds). Our loop is there for that case. We've seen it place one hold. We haven't seen it place two. ### The trigger fired on test-gateway orders We weren't sure it would. The run list shows four completed runs: three with all three actions, and one with none, which was #1035. So Order risk analyzed ran even though every risk panel said "Analysis not available". That's one store, and Shopify doesn't document it. ### The daily digest A held order nobody looks at is just late. The second workflow, "Fraud hold: daily digest", runs at 08:00 store time, finds orders tagged fraud-hold, oldest first, and emails the reviewer one line per order: name, created time, total, Shopify's recommendation and an admin link. It closes with a reminder to swap the tag after each decision. Its test run listed #1034, #1036 and #1037 and nothing else. The first scheduled run was due on 22 October, so what you see is Flow's preview, not an inbox. ### Releasing a held order by hand Flow has a release action, but Shopify says it releases all fulfillment holds on the fulfillment order (Shopify Help Center, Release fulfillment order holds), other apps' holds included. So a person decides after checking the order: release, contact the customer, or cancel and refund. Our customer was fictional, so we went straight to the mechanics with #1036. We clicked Release hold. Shopify showed "Fulfillment released", the order went back to Unfulfilled, and the timeline recorded "You released a hold placed by Shopify Flow on 1 item." Then we swapped the fraud-hold tag for fraud-reviewed. Filtering the orders list on fraud-hold, the digest's own query, then returned #1034 and #1037 only. We haven't watched a scheduled digest drop #1036, but the search it depends on already has. The tag and the timeline entry also show that someone looked before the order shipped. That is a record for the store, and we make no claim about how a bank weighs it. ### What we could not show - Shopify's own risk signals. They are in the rule, but no test order here had a risk result to match, because simulating one needs Shopify Payments test mode. On a client store with Shopify Payments, the first acceptance test covers them with Shopify's own test values. - Those signals under the threshold. #1035 showed the new-customer pair is grouped correctly. An order Shopify flags below $200 couldn't be placed here. - Manual capture. The optional routine that holds back the payment is untested live. Shopify Payments gives a 7-day authorisation window and charges 1.75% on top of the usual fees to capture later (Shopify Help Center, payment authorization), so the handover is clear about the clock. - A split order. One fulfilment order held, two not yet seen. - Digital or auto-fulfilled items. No Shopify page says whether they can be fulfilled before a hold lands, so we treat them as out of scope until tested on the store. - Email delivery. On the hold workflow, Flow's run detail shows the Send internal email step as Completed. The digest has only run as a test event, which shows a preview and sends nothing. We read both from Flow, not from an inbox, so we haven't shown an email arriving. ### Three things that caught us out The demo hold email is shorter than the one in the kit, which says which signal matched and links to the order. We also changed the hold note after the first four orders, to the wording the kit uses for clients: "Fraud hold (Axioprax KIT-04): held for fraud review. Release by hand after checking." A fifth test order, #1038, a new customer at $749.95, was held with the new note and tagged fraud-hold. Time zones bit twice. The digest preview shows #1034 created at 14:25, while the admin shows 10:25 am. Four hours is the gap between UTC and New York in October, so we read it as the digest printing UTC, and the client version labels the time zone. And 08:00 store time here is 17:30 in India, so for a client we set the schedule from where the reviewer sits. Then there are orders Shopify doesn't analyse. Test, free, POS, B2B and subscription renewal orders get no real recommendation (Shopify Help Center, fraud analysis), and no page covers draft orders marked as paid. Our test orders still fired the trigger, but we haven't seen which of the other types do, so we test the ones a client's store uses. ### What the kit leaves out It doesn't win or manage disputes, and no rule cancels or refunds an order without a person approving it. The rest, from payout holds to third-party fraud apps, is on the kit page (/store-kits/fraud-hold). ### Getting started The kit needs someone at the store who can check held orders every working day. Without that person it just makes orders late. If you have them, and risky orders are shipping before anyone looks, see what is in the KIT-04 Fraud Hold kit (/store-kits/fraud-hold), then describe your store at axioprax.com/start (/start?kit=fraud-hold). Tell us your payment gateway, daily order volume and what you've been seeing. Black Friday is on 27 November this year. We confirm the kit fits before you pay. ### Frequently asked questions ### Will this stop chargebacks? No promise of that. The kit makes sure a person checks a risky order before it ships and keeps the evidence banks ask for ready. Banks decide disputes, and we don't claim any reduction. ### Do I need Shopify Payments or Shopify Plus? Neither, strictly. Flow is free on Basic, Grow, Advanced and Plus. Shopify's fraud recommendation is fullest on Shopify Payments, and on other gateways it depends on your plan and processor, so on those the rule leans more on your own signals. Our demo store is a Plus development store. ### What happens to orders Shopify doesn't analyse? Shopify's own signals can't match them, because Shopify says free, POS, B2B and subscription renewal orders get no real recommendation. On our demo store, test orders still reached the rule and the store's own signals held them. We test the order types your store uses and record in the handover which ones reach the rule. ### What this article is built on Shopify claims were read against Shopify's own pages, and demo results come from our build log and numbered captures, all on 21 October 2026. Claim ledger. Demo results are from a demo store with synthetic data, not a client. #ClaimSourceCheckedStatus 1Order risk analyzed starts when fraud analysis is complete; third-party assessments don't fire itShopify Help Center, Order risk analyzed21 Oct 2026Verified 2Recommendation (accept, investigate, cancel) and risk level are readable in Flowshopify.dev, OrderRiskSummary and enums21 Oct 2026Verified 3Hold fulfillment order places a hold per fulfilment order; release action removes all holds, including other apps'Shopify Help Center, Hold fulfillment order; Release fulfillment order holds21 Oct 2026Verified 4No Shopify risk template holds fulfilment; capture templates need manual captureShopify Help Center, Flow templates for orders21 Oct 2026Verified 5Some card-testing attempts might result in placed ordersShopify Help Center, preventing fraud21 Oct 2026Verified 6Fraud dispute evidence; deadline typically 7 to 21 days; US fee $15, returned if won, varies by countryShopify Help Center, chargeback process; resolve a chargeback21 Oct 2026Verified 7Simulated risk on test orders needs Shopify Payments test modeShopify Help Center, fraud analysis21 Oct 2026Verified 87-day authorisation on Shopify Payments; 1.75% to capture laterShopify Help Center, payment authorization21 Oct 2026Verified 9A fraud order following failed payments classed as low riskShopify Community thread 672811, posted 27 Aug 202621 Oct 2026Reported by a merchant 10#1034 held and tagged; #1035 not held; #1036 held on country mismatch; #1037 held with one fulfilment orderDemo store build log, captures 03, 05, 06, 1021 Oct 2026Observed 11Order risk analyzed fired for test-gateway orders showing "Analysis not available"Demo store run list, capture 06, 1221 Oct 2026Observed on one store 12Digest test run listed #1034, #1036, #1037Flow test event preview, capture 0821 Oct 2026Observed (preview, not inbox) 13Release by hand, tag swap, fraud-hold search returns #1034 and #1037 onlyCaptures 11, 12, 1321 Oct 2026Observed 14New hold note in the workflow; #1038 held with it and tagged fraud-holdCaptures 16, 1721 Oct 2026Observed 15Shopify's recommendation and risk level as hold signals; manual capture; split-location hold; scheduled digest run; email deliveryNot shown on the demo store21 Oct 2026Not yet proven ### Shopify Fake Sign-ups: Bot Cleanup on a Demo Store https://www.axioprax.com/insights/bot-cleanup-demo-build (2026-10-20) Shopify now hides bots from session reports, but they stay in your list and your bill. How we tagged fake sign-ups and guarded abandoned-checkout emails. Shopify's September 2026 analytics update filters identified bot sessions out of session reports by default. It doesn't block them. Shopify says bot filtering applies only to sessions-related metrics, and its changelog says orders, sales and customer counts aren't changed, so fake sign-ups are still in your customer list, still on your email bill and still able to start abandoned-checkout emails. This is how we cleaned that up on our own demo store with Shopify Flow: a tag on suspected bot accounts, a guard on the abandoned-checkout email, and a safety net for real buyers. It is for merchants whose customer list or email bill is growing faster than their sales. Our kit, KIT-02 Bot Cleanup, does not block bots, and nobody outside Shopify can. What the kit includes and its price are on the KIT-02 Bot Cleanup page (/store-kits/bot-cleanup). This is a demo build, not a client project. Everything below was built and tested on Northfield Outfitters, our own Shopify development store: demo store, fictional shop, synthetic data, real tools. There are no client results in this article. ### Key points - Shopify's bot filtering applies only to session metrics. It hides identified bots from session reports and does not block them (Shopify Help Center, bot filtering, checked 20 October 2026). - Klaviyo doesn't bill suppressed profiles, Mailchimp still bills unsubscribed contacts, and Shopify Messaging bills per email sent, so the fix depends on the email tool (each vendor's help pages). - On our demo store, a Flow workflow on Customer created tagged 17 of 17 imported fake accounts and none of the 4 genuine-style or 19 existing customers. - The kit's abandoned-checkout guard is tag-only: no reminder for an account tagged suspected-bot. On the demo store a tagged test customer with a full address got no reminder step, and an untagged one with no address went on to it. - A third workflow takes the tags off anyone who places an order. A tagged test customer's $1 order cleared both tags. - We never delete. You decide. Axioprax tags, suppresses or archives only after the owner approves the list, and anyone with an order is protected. ### What did Shopify's bot filtering change? Your session reports. Nothing else. The rollout ran from 21 to 23 September 2026, and once it reached a store, identified bot sessions are filtered out of session-related reports by default (Shopify Help Center, bot filtering). You can bring them back one report at a time with the "Human or bot" filter. Sessions from before 7 October 2025 aren't classified at all. Shopify's changelog post on the update says Orders, sales, and customer counts aren't changed and asks merchants to treat the new figures as a new baseline (Shopify changelog, session measurement improvements). Conversion rates look more believable now. The fake accounts haven't gone anywhere. ### Why doesn't Shopify just block the bots? Its help page says blocking bots from a site entirely has drawbacks, so Shopify's approach is to display and filter bot traffic in analytics instead (same page). We don't put a firewall in front of Shopify for clients either. Merchants on the Shopify Community describe what is left. One Plus merchant reported abandoned checkouts rising from about 130 in August to more than 11,000 after 1 September (Shopify Community thread 684931). Another thread counts 2,066 bot accounts (thread 607939). Those are individual merchants' reports, not a benchmark. ### Where do bots cost a Shopify store money? Mostly in the email tool, and each tool counts differently. Klaviyo bills on profiles it can email, and suppressed profiles don't count towards the plan (Klaviyo Help). Mailchimp is the one that catches people out: subscribed, unsubscribed and non-subscribed contacts all count, and only archived, cleaned or deleted contacts stop counting (Mailchimp Help). Unsubscribing a bot there saves nothing. Archiving does, though Mailchimp notes that archiving isn't a legal substitute for deletion. Shopify Messaging, which used to be Shopify Email, bills per email sent: 10,000 free a month, then $1 per 1,000 up to 300,000 emails (Shopify Help Center, Shopify Messaging pricing). Abandoned-checkout automation emails don't count towards that. So for a Messaging store under the free allowance the saving is $0, and the gain is cleaner open rates. ### What does the Bot Cleanup kit include? Six parts. The damage report comes first, because it shows what the rest is worth on your own plan. - Bot damage report. One page in plain English: how many accounts look automated over the last 90 days, where the fake traffic comes from, and what it costs on your own email plan. - Sign-up form check. hCaptcha and Shopify's marketing double opt-in, or the equivalent settings in Klaviyo or Mailchimp. - Scoring workbook and approved list. Every customer scored on six signals. Anyone with an order is marked PROTECTED and can't be flagged. You approve the list in writing before anything changes. - Suppression in your email tool. Suppress in Klaviyo, archive in Mailchimp, or leave out of Shopify Messaging sends. We never delete customers or profiles. - Shopify Flow workflows. One tags new suspected bots, one keeps abandoned-checkout reminders away from tagged accounts, and a third removes the tags if a tagged customer buys. - Handover. Session reports shown with bots hidden and included, and a note on reviewing the suspected-bot tag each week. If you are looking at Flow for other store jobs as well, our guide to Shopify automation workflows (/insights/shopify-automation-workflows-guide) covers more of them. ### What did we build it on? Northfield Outfitters, a fictional shop on a Shopify Plus development store of ours. Before we started it had 19 customers, 16 of them with orders, no email app and no Flow. We installed Flow and it ran. Shopify's development store page only mentions Flow for Plus, so we can't say whether it runs on other kinds. We then wrote a 25-row customer file on reserved test domains: 21 fakes in four patterns (random addresses on three throwaway-style domains, and a link in the name) and 4 genuine-style customers, one with capital letters in the address and one with a plus sign in it. ### Were the sign-up forms already protected? Partly. hCaptcha was already on for contact and comment forms and for login, create account and password recovery pages, which matches Shopify's note that it is on by default (Shopify Help Center, online store preferences). Marketing double opt-in was off. We turned it on under Settings, Notifications, Customer notifications. Shopify's documents disagree about whether hCaptcha covers the newsletter form. The test theme on our demo store has no newsletter form at all, so we couldn't check it there. The kit tests that form on each client's store before relying on it. ### How do you tag suspected bot customers in Shopify Flow? With a workflow on the Customer created trigger. Ours checks that the customer has 0 orders, then checks whether the email ends with one of the listed throwaway domains or the first or last name includes "http". If either is true, Flow adds the tag suspected-bot. The email check uses the customer's default email address field, because Shopify lists the older email field as deprecated (shopify.dev, Customer object). We drafted it with Flow's own assistant from our written spec and checked each step against the spec before turning it on. The tag is a flag for a person to review, never a verdict. ### Does Customer created fire on a CSV import? On our store it did, although Shopify's help page only mentions new orders and profiles created by hand. We imported the 25 rows with overwrite off. Shopify brought in 21 and refused 4, with 0 duplicates, and the 4 it refused were exactly the rows with a link in the name. Flow ran on the import. All 17 fakes that got in were tagged suspected-bot. None of the 4 genuine-style customers was tagged, and neither was any of the 19 customers already on the store. Because Shopify refused the link-in-name rows, the "http" part of our condition never met a real record. Flow also ran each time one of our test checkouts created a customer, judging by the run times. We haven't tested storefront account sign-ups: the store uses Shopify's newer customer accounts, and creating one is the owner's job, so the kit checks it on each client's store. ### Is the Ends with check case-sensitive? Not on our store. We created a customer by hand as KQ7X9Z2W4V@SIGNUP-BOT.TEST with 0 orders. Shopify kept the address in capitals, as typed, and Flow still matched it and tagged it. ### What about the bots already in the list? Flow only sees new customers, so existing junk goes through the scoring workbook. It scores every row on six signals, and anyone with an order is PROTECTED whatever the address looks like. Run on the 25-row demo file, it flagged 21 of 21 fakes (scores 7 to 10) and 0 of 4 genuine-style customers, who all scored 1 or 2 and read OK. In our 24-row edge-case file, a real buyer on a throwaway-looking domain stayed PROTECTED. A burst of nine real newsletter sign-ups within 35 minutes wasn't flagged: seven read OK, and the two with a birth year in the address landed in REVIEW for a person to check. Three bot sign-ups on a test domain within three minutes were still flagged. Bulk tagging goes one listed domain at a time, and on a client's store only after the list is approved. In the segment editor we ran customer_email_domain = 'signup-bot.test' AND number_of_orders = 0 and got 9 customers: the 8 imported fakes on that domain and our capital-letters test account. We selected all 9 and added the reviewed tag in one step. Then came the check the kit never skips: two segments for tagged customers with at least one order. Both returned 0. We also ran the demo store's own customer export through the workbook, in both of Shopify's CSV formats. The results matched row for row: all 17 customers with an order were PROTECTED, all 17 imported fakes and the capital-letters test account were flagged, and a test account we added later on a listed domain, with an ordinary-looking name, landed in REVIEW for a person to check. REVIEW means a person looks. It never means a quiet approval: the workbook asks for a written reason before a REVIEW row can go on the approved list, and it refuses anyone with an order outright. Nothing gets tagged, suppressed or archived from the workbook until the store owner approves the list in writing. Axioprax never deletes customers or data. If an owner wants records removed, that is their decision and they do it by hand. Shopify says a deleted customer profile can't be restored, and customers with orders can't be deleted at all (Shopify Help Center, manage customers). ### How do you stop abandoned-checkout emails going to bots? Start with what Shopify already does. Checkouts it identifies as suspected card testing or bot activity don't create abandoned checkouts in the first place (Shopify Help Center, Customer abandons checkout trigger), and its abandoned-checkout automation sends only to email marketing subscribers unless you change it. We installed Shopify Messaging and opened its Recover abandoned checkout automation in Flow. It held more checks than we expected. It skips anyone who abandoned another checkout in the last 24 hours, anyone who has ordered since, and checkouts whose items are no longer in stock. The kit's guard goes just before the send step, and it is a single condition: none of the customer's tags equal suspected-bot. An account carrying the tag gets no reminder. Everyone else carries on through Shopify's own checks as before. ### Why is the guard tag-only by default? Because the obvious extra check costs real reminders. Junk checkouts often stop after the email field, so it is tempting to send a reminder only when a shipping address has been filled in. Flow can do that. Its condition editor offered a "Not empty and exists" operator for the address line, which Shopify's conditions page doesn't list, so we tested it before trusting it. The trouble is who else has no address. A real shopper who types an email and leaves before the address step gets no reminder, and that is exactly the path our test A took. Checkouts for local pickup, digital-only products or services may not ask for an address at all, so those shoppers would lose their reminder too. Shopify's help pages on pickup and digital products don't say whether checkout asks for an address, and we haven't run those checkouts through the condition. So in the kit the address condition is optional. We add it only if the store owner approves it in writing after we have explained that trade-off, and we only suggest it for a store that ships every order and sees a lot of email-only junk checkouts. The tag-only guard has a smaller worst case. A real person who was tagged by mistake misses a reminder until the weekly review of the suspected-bot tag takes the tag off. ### Did the guard work? Yes, both ways, in two rounds of tests. The first round, on 18 October, ran an earlier design that also needed a shipping address. Test A had an email and no address and stopped at the guard. Test B had a made-up Denver address and reached the send step. Both test customers were untagged, so those runs proved the address condition and said nothing about the tag. That gap is why we went back. On 20 October we switched the demo store to the tag-only guard and shortened the wait to six minutes for the test. Test C was a customer of our own, tagged suspected-bot and subscribed, who started a checkout for a $699.95 board with a full address and left. Every earlier check came out true, the guard did not, and the run finished with no results: Send marketing email never ran. Test D was the control: untagged, an email and nothing else. Its run passed the guard and went on to the send step. Then we set the wait back to Shopify's 10 hours. Test D's run tells you something about the address condition too. With the tag-only guard, a shopper who leaves after typing an email still gets to the reminder step, which is the point of keeping that condition optional. What we haven't confirmed is delivery. Flow recorded the send step, but we haven't checked that a development store's reminder actually lands in an inbox, so we read every result from the run log. ### What if a real customer gets tagged? A third workflow takes the tag off. It runs on Order created, reads the customer's tags (not the order's, which is an easy field to pick by mistake), and removes both suspected-bot and the reviewed tag in one step, because anyone who buys counts as a real customer in this kit. We tested it with a customer of our own carrying both tags. A $1 draft order, marked as paid, was enough: when we opened the customer again, only our test label was left. Shopify won't delete a customer who has an order, and we don't delete records anyway, so that test customer stays on the demo store. That is why the kit only runs this test on a client's store with their written OK. One case we couldn't test: a brand-new customer on a listed domain who buys in the same moment they sign up, where both workflows fire at once. Shopify refused a draft order for our reserved test domain. The weekly review and the two protection segments catch that case either way. ### What caught us out? A few things in testing, all now in the runbook. The storefront kept one checkout per browser, so starting test B at once would have overwritten test A. We waited until A showed as abandoned, which took longer than ten minutes, then cleared the cart. Shopify's 24-hour rule meant two test checkouts from one customer wouldn't both send, which is why A and B used different addresses. Double opt-in was another. Test A's customer record showed "Not subscribed" even though the box was ticked. We think the address was waiting for its confirmation email, but we haven't checked, so we read results from the Flow run log rather than an inbox. The scoring workbook doesn't work in Google Sheets. Opened there, its protection check reads FAIL with a formula error and one sample row scores differently. It fails safe, so nobody can tag from a broken copy, but the kit uses LibreOffice only. Shopify emailed our customer export instead of downloading it, with 46 customers. Its help page says exports of more than 50 are emailed, so plan for an email either way. ### What the kit does not do - It doesn't block bots. Nobody outside Shopify can stop traffic reaching a Shopify-hosted checkout. The kit cuts what bots cost you and how much they cloud your reports. - No Cloudflare or firewall set-up in front of Shopify. - No deleting. Axioprax never deletes customers, profiles or data. Removing records is the owner's own decision, done by hand, and Shopify won't delete anyone with an order anyway. - No ad-account clean-up inside Meta or Google beyond leaving bot sources out of reporting. - No shipping-address filter in Klaviyo. Its Started Checkout flow filters can only use top-level event properties, and shipping address isn't one of them (Klaviyo Help), so we use a subscriber and profile filter there instead. - The demo used Shopify Messaging only. We haven't run the Klaviyo or Mailchimp steps on the demo store. ### Is the kit a fit for your store? - You are seeing fake customer accounts, junk abandoned checkouts, odd traffic spikes, or an email bill rising faster than sales. - Your email tool is Klaviyo, Mailchimp or Shopify Messaging. Others get a custom quote. - Your plan includes Shopify Flow, which Shopify lists as free on Basic, Grow, Advanced and Plus. - Someone at the store can approve the list of suspected accounts in writing. We work through a Shopify staff or collaborator account and an email-tool user you invite, never an owner login, and no passwords by email. The kit is $349. ### Getting started If bots are filling your customer list or your email bill, see what is in the KIT-02 Bot Cleanup kit (/store-kits/bot-cleanup), then describe your store at axioprax.com/start (/start?kit=bot-cleanup). Tell us which email tool you use, roughly how many customers you have and what you are seeing. We confirm eligibility before you pay. ### Frequently asked questions ### Does Shopify's bot filtering block bots? No. It filters identified bot sessions out of session reports by default, and you can show them again per report with the "Human or bot" filter. Orders, sales and customer counts aren't changed, so fake accounts stay in your customer list. ### Can you block bots from my Shopify checkout? No. Nobody outside Shopify can stop traffic reaching a Shopify-hosted checkout. We clean fake sign-ups out of your email list, tag suspected bot accounts for review and stop abandoned-checkout emails going to junk checkouts. ### Will cleaning up bots lower my email bill? It depends on the tool. Klaviyo doesn't bill suppressed profiles, and Mailchimp stops counting archived contacts but still counts unsubscribed ones. Shopify Messaging bills per email sent, so below its free monthly allowance the saving is $0. Your bot damage report shows the figure from your own plan. ### Will you delete the fake customers? No. Axioprax never deletes customers or data. We tag, suppress or archive them after you approve the list. If you want records removed, that is your decision and you do it by hand. Shopify says deleting a customer profile can't be undone, and customers with orders can't be deleted. Anyone with an order is protected and never flagged. ### Will the abandoned-checkout guard stop reminders to real shoppers? By default it only stops them for accounts tagged suspected-bot, so the worst case is a wrongly tagged person missing a reminder until the weekly review clears the tag. A shipping-address condition is optional, because it also stops reminders for shoppers who leave before the address step and for checkouts that may not ask for an address, such as local pickup, digital-only or service products. We add it only with your written approval. ### Does it work with Klaviyo? Yes, for suppressing approved junk profiles and filtering the abandoned-checkout flow to real subscribers. Klaviyo's Started Checkout filters can't see a shipping address, so the optional shipping-address condition is only possible in Shopify's own abandoned-checkout automation. ### Do I need Shopify Plus? No. Shopify lists Flow as a free app on Basic, Grow, Advanced and Plus. Our demo store is a Plus development store, and we check that Flow is installed on your store during the access check. ### What this article is built on Every vendor claim was read against its own page on 16 October 2026, and the claims about Shopify, Klaviyo and Mailchimp that we also make on the kit page were read again on 20 October 2026. Each community report is labelled as a report. Every demo result comes from our build log and captures for the demo store, recorded on 18 and 20 October 2026, with synthetic data and test customers of our own. Claim ledger. Demo results are from a demo store with synthetic data, not a client. #ClaimSourceCheckedStatus 1Bot filtering applies only to session metrics; Shopify filters bots from analytics rather than blocking themShopify Help Center, bot filtering20 Oct 2026Verified 2Rollout 21 to 23 September 2026; bot sessions filtered by default; per-report "Human or bot" filter; no classification before 7 October 2025Same page20 Oct 2026Verified 3Orders, sales and customer counts aren't changedShopify changelog, session measurement improvements20 Oct 2026Verified 4Abandoned checkouts from about 130 to more than 11,000; 2,066 bot accountsShopify Community, threads 684931 and 60793918 Oct 2026Reported by merchants 5Klaviyo doesn't bill suppressed profiles; Mailchimp bills unsubscribed contacts, not archived onesKlaviyo Help; Mailchimp Help, pricing plans20 Oct 2026Verified 6Shopify Messaging bills per email sent: 10,000 free a month, then $1 per 1,000 up to 300,000; abandoned-checkout automation emails don't countShopify Help Center, Shopify Messaging pricing20 Oct 2026Verified 7Suspected bot and card-testing checkouts don't create abandoned checkouts; the automation sends to subscribers by defaultShopify Help Center, Customer abandons checkout; migrating abandoned checkout20 Oct 2026Verified 8Customer created is documented for new orders and manual creation; it also fired on CSV importShopify Help Center, Customer created; demo store18 Oct 2026Verified; import observed 917 of 17 imported fakes tagged; 0 of 4 genuine-style and 0 of 19 existing; 4 link-in-name rows refused at importDemo store build log, synthetic data18 Oct 2026Observed 10Customer email field deprecated in favour of the default email addressshopify.dev, Customer object16 Oct 2026Verified 11Deleting a customer can't be undone; customers with orders can't be deletedShopify Help Center, manage customers16 Oct 2026Verified 12Klaviyo Started Checkout filters use top-level properties only, with no shipping addressKlaviyo Help16 Oct 2026Partly verified: no shipping address listed among the filter properties 13Optional address-plus-tag guard: test A (no address) stopped before the send step; test B's Send marketing email step completed. Both test customers were untaggedDemo store Flow run log18 Oct 2026Observed 14Tag-only guard: test C (tagged, full address) stopped before the send step; test D (untagged, no address) reached itDemo store Flow run log20 Oct 2026Observed 15Pickup and digital-only checkouts carry no shipping addressShopify Help pickup and digital pages don't say; only relevant to the optional address condition20 Oct 2026Not tested 16Scoring workbook: 21 of 21 demo fakes flagged, 0 of 4 genuine-style; edge file buyer PROTECTED, a 9-sign-up real burst seven OK and two REVIEW, a 3-bot burst flaggedKIT-02 DEMO-IMPORT and EDGE-CASES workbooks, synthetic data18 Oct 2026Observed 17Reminder email delivered from a development storeNot confirmed; results read from the Flow run log20 Oct 2026Not yet proven 18Safety net removes the tags when a tagged customer orders; storefront sign-up and the newsletter formDemo store test order (safety net); sign-up and newsletter form not testable on the demo store20 Oct 2026Safety net observed; sign-up not tested 19Flow runs on a Plus development storeDemo store; other development stores not documented18 Oct 2026Observed, Plus only 20Domain segment returned 9; both protection segments returned 0; real export in both CSV formats gave identical scores, every customer with an order PROTECTEDDemo store segments; KIT-02 loader20 Oct 2026Observed 21Workbook in Google Sheets: protection check FAIL, one row scored differentlyDemo copy in Google Sheets20 Oct 2026Observed 22Customer created also runs for customers made by a storefront checkoutDemo store Flow run times; the run list does not name the customer20 Oct 2026Inferred from timing ### 7 Shopify Automations Worth Building First https://www.axioprax.com/insights/shopify-automation-workflows-guide (2026-08-28) Abandoned carts, stock alerts, post-purchase emails, VIP tagging, returns and weekly reports: seven Shopify automations worth building first, and why. Most stores don't need a big automation project to stop the busywork. They need seven small jobs to run on their own, and most of those can be built with tools that are already in the Shopify admin. Here they are in the order we'd build them, with the tool for each, the detail that usually gets missed, and the point where a free setup runs out. ### 1. Abandoned checkout messages When someone fills in checkout and leaves, a reminder is the most common store automation and the easiest to switch on. Shopify Messaging has ready-made automations for abandoned checkouts, abandoned carts and browse abandonment, and in May 2026 Shopify added SMS versions of them (Shopify changelog (https://changelog.shopify.com/posts/create-sms-marketing-automations-in-shopify-messaging)). Shopify Flow also has a Customer abandons checkout trigger, and its Recover abandoned checkout template sends an email 10 hours after a customer reaches checkout without ordering (Shopify Help (https://help.shopify.com/en/manual/shopify-flow/reference/triggers/customer-abandons-checkout)). Two things matter more than the wording. Send only to customers who agreed to marketing, which the Flow template does by default. And give each checkout one owner: if Klaviyo or another email app already sends a reminder, switch off the Shopify one, or customers get two. Shopify says checkouts it flags as card testing or bot activity don't create abandoned checkouts at all. Some fake sign-ups still slip through, which is what our Bot Cleanup demo build (/insights/bot-cleanup-demo-build) deals with. ### 2. Order and shipping updates These are built in, which is exactly why they get forgotten. Shopify sends an order confirmation and a shipping confirmation, and where the carrier reports it, "out for delivery" and "delivered" emails as well. Those last two can be switched off in the admin (Shopify changelog (https://changelog.shopify.com/posts/changes-to-shipping-delivery-notifications)), so check they're on. Then open each notification under Settings and read it as a customer would: the wording, the tracking link, the reply-to address. What the built-in emails won't do is notice a parcel that has stopped moving. If "where is my order" messages fill a lot of your inbox, that's the point to add tracking alerts on top. ### 3. Low-stock alerts Flow includes a template that tells you when a product's inventory is low. It runs on the Inventory quantity changed trigger, and Shopify's own advice is to check both the new quantity and the previous one, so the alert fires once when stock crosses your line instead of on every order after it (Shopify Help (https://help.shopify.com/manual/shopify-plus/flow/examples)). A fixed number is a blunt tool, though. Ten units is a month of stock for one product and a day for another. If you sell fast-moving lines, the next step is an alert based on days of stock left, which needs your recent sales as well as your stock. Our guide to n8n e-commerce automations (/insights/n8n-ecommerce-automation-workflows) shows that version. ### 4. Customer tiers by spend Flow's "Organize customers by lifetime spend tiers" template tags customers as their total spend passes levels you set. Once the tag exists, your email tool can treat those customers differently: earlier access to a launch, a thank-you note from the founder, a different offer. Keep the tiers few and the thresholds round, so anyone on the team can explain them. And only use a tag for treatment you'd be comfortable describing to the customer who asked why they got it. ### 5. Review requests after delivery A review request that lands before the parcel does is wasted. Most review apps can send their own request and time it from fulfilment or delivery, so use that rather than building one from scratch. Two checks make the requests better. Skip orders with an open return or refund, and skip orders that arrived late. Both groups have something to tell you, but a review form isn't the place to hear it. ### 6. Returns With self-serve returns, customers request a return or a cancellation from their order status page instead of emailing you. Your return rules decide which items qualify, how long customers have and who pays for return shipping. You still approve each request, then add any exchange items and send the return label (Shopify Help (https://help.shopify.com/en/manual/orders/self-serve-returns)). That removes most of the back-and-forth. Damaged items, disputes and anything outside the rules stay with a person. ### 7. A weekly sales summary Shopify doesn't email reports on a schedule by itself. Flow now gets close. Its Scheduled time trigger can run weekly, and the Get analytics data action runs a ShopifyQL query and hands the numbers to the next step, such as an internal email (Shopify Help (https://help.shopify.com/en/manual/shopify-flow/advanced-workflows), Get analytics data (https://help.shopify.com/en/manual/shopify-flow/reference/actions/get-analytics-data)). It sends values, not a formatted file, so keep it to five or six numbers: sales, orders, average order value, refunds and your top products. If you want a comparison with the same week last year, charts, or figures from outside Shopify such as ad spend, that's a job for n8n or a reporting app. ### What it costs Flow is a free app on the Basic, Grow, Advanced and Plus plans. Its Send HTTP request action, which lets a workflow call other apps, needs Grow or above (Shopify Help (https://help.shopify.com/en/manual/shopify-flow)). Shopify notifications and self-serve returns come with the admin. SMS in Shopify Messaging is billed, and you set a spending limit in its settings. So most of these cost setup time rather than a subscription. Each one still needs testing with a real order before you trust it, and a look every few months to check it still runs. ### Where to start Go down the list and mark which of the seven you already have. Turn on the built-in ones first, the notifications and returns, then build the Flow ones in order. If you'd rather have someone go through the store with you, our Store Ops Audit checks what's already running and ranks what to fix first. ### Building an AI Chatbot for Your Shopify Store https://www.axioprax.com/insights/shopify-ai-chatbot-guide (2026-08-28) An AI chatbot trained on your products and policies can answer routine customer questions. How to build one for a Shopify store that hands over when it should. Look through a month of customer messages in most Shopify stores and the same handful of questions come up again and again. Where is my order? Can I return this? Do you ship to my country? Which size should I get? Nearly all of them can be answered from two places: things you've already written down, and the customer's own order. That's what a chatbot for a Shopify store has to do well. Here's what it should answer and what it should pass to a person, the three ways to add one today, and how to handle order questions without giving away anyone's details. ### What it should answer, and what it should hand over A good store chatbot answers questions about your products, your policies and the customer's order, and says plainly when it doesn't know. It hands over anything that needs a decision: a damaged item, a refund outside your rules, an angry customer, a wholesale enquiry. ### Three ways to add one Shopify Inbox. Shopify's own chat app is listed as free, and you can run it with or without its Inbox agent (Shopify App Store (https://apps.shopify.com/chat)). Turned on, the agent answers questions about your products, shipping, returns, sizing and store policies from your catalog, your policies and your knowledge base content, and your conversations stay with your staff until you turn it on (Shopify Help Center, Shopify Inbox (https://help.shopify.com/en/manual/inbox/ai)). Since October 2026 that knowledge base is built into the Shopify admin, where you review and publish the facts AI agents get about your store (Shopify Changelog, 26 October 2026 (https://changelog.shopify.com/)). Inbox also has a default Track my order answer that uses an order number and email. It's the place to start if most of your questions are simple and you want a person to stay close to the conversation. A chatbot app. Dedicated apps give you more control over what the bot knows and how it hands over, usually for a monthly fee. Before you pick one, check three things: can it read order status, can it hand a conversation to your email or helpdesk with the history attached, and can you see every conversation afterwards? A custom build. Shopify now publishes Storefront MCP, which lets an AI assistant search your catalogue, build a cart and answer questions about your policies, shipping and returns. A separate customer accounts server is for tracking orders and managing returns (Shopify.dev (https://shopify.dev/docs/apps/build/storefront-mcp/index)). A custom build makes sense when you need answers grounded only in your own data, with your rules for what it may and may not do. ### Write the knowledge down first Whichever route you take, the bot can only be as good as what you give it. Before you switch anything on, collect your shipping and returns policies, your FAQ, size guides and product details, and the twenty questions your team answers most. Then read them as a stranger would. Look for gaps and contradictions. If your returns page says 30 days and an old FAQ says 14, the bot will quote whichever it finds, and the customer will hold you to it. Date your policies, delete the old versions, and write answers in full sentences rather than notes. Our guide to training a chatbot on your business (/insights/train-ai-chatbot-business-knowledge) goes through this step by step. ### Order status without leaking orders "Where is my order?" is usually the most common question, and the one most often handled badly. An order number on its own is not proof of who's asking. Order numbers are short, often sequential, and printed on packing slips and in forwarded emails. Either the customer is signed in and sees only their own orders, or they give the order number and the email used at checkout and both must match. If they don't match, the bot says it can't find the order and offers a person. It shouldn't hint at why. ### Make the handover smooth Customers don't mind being passed to a person. They mind repeating themselves. When the bot hands over, the conversation so far should go with it, and the customer should be told when to expect a reply. If nobody is online, say so and give the hours, rather than leaving them waiting in a chat window. ### Measure it with real conversations Before launch, take fifty real questions from your inbox and run them through the bot. Mark each answer right, wrong or handed over, and fix the knowledge behind every wrong one before going live. After launch, watch three things: how many conversations end without a person, how many get handed over, and how many answers were wrong when someone reads the transcripts. Read a sample every week for the first month. Most wrong answers trace back to something missing or out of date in the knowledge, not to the bot itself. If you're still deciding whether a bot is right for your store at all, our comparison of AI chatbots and live chat (/insights/ai-chatbot-vs-live-chat) covers when each one fits. ### E-commerce Automation ROI: The Payback Maths and What to Automate First https://www.axioprax.com/insights/ecommerce-automation-roi (2026-08-29) Work out whether an automation pays for itself in ten minutes with your own numbers, see where the revenue side is real, and pick what to automate first in an online store. Every automation pitch promises a big return. Very few show the arithmetic. The good news is that you don't need a consultant to work out whether an automation is worth building. You need about ten minutes, a calculator, and honest numbers from your own store. This guide walks through that calculation, where the revenue side is real and where it's guesswork, how to decide what to automate first, and the cases where automation isn't worth it at all. ### The calculation An automation earns its keep in three ways: it saves time, it prevents mistakes, and sometimes it brings in revenue. Against that you set what it costs to run each month. Then divide the cost of building it by what it gains you each month, and you know how long it takes to pay for itself. Here's a worked example. Say someone on your team spends 15 minutes on each order: checking it, tagging it, copying details into the shipping tool and a spreadsheet. At 200 orders a month, that's 50 hours. For the hourly cost, the US Bureau of Labor Statistics puts the median wage for customer service representatives at $21.53 an hour in May 2025 (BLS (https://www.bls.gov/ooh/office-and-administrative-support/customer-service-representatives.htm)). That's wages only, before benefits or overheads, so it's a conservative figure. 50 hours at $21.53 is about $1,076 a month. Take off running costs, say $50 a month for the tools the automation uses, and you're gaining about $1,026 a month. A build that costs $2,500 pays for itself in under three months. If it costs $10,000, it takes nearly ten, which may still be worth it, but now it's a real decision rather than an obvious one. Two things to watch. Use your own numbers, not ours. And time saved only counts if that person does something more useful with it. If the 50 hours just disappear into the week, the saving is real but nobody will feel it. ### Where the revenue side is real Revenue lift is where most ROI claims get wishful. The one area with solid evidence is automated email. Klaviyo's 2026 benchmarks, drawn from more than 183,000 of its customers, found that automated flows such as welcome, abandoned cart and post-purchase emails generate nearly 41% of email revenue from just 5.3% of sends (Klaviyo (https://www.klaviyo.com/products/email-marketing/benchmarks)). That's vendor data, so treat it as a strong signal rather than a promise, but it matches what most store owners see once those flows are switched on. Elsewhere, count revenue only when you can measure it. A review request that goes out after delivery can be measured in reviews collected. "Better customer experience" can't, so leave it out of the sum. ### What to automate first One question tells you whether a process is ready to be automated: can you write down every step on paper? Not a rough description. Every step, including what happens when something goes wrong. If you can't, the automation will be built on guesses, and guesses break. After that, look for four things together: - Repetitive. It happens the same way every time. - Rule-based. The decisions follow clear rules, not judgement. - High volume. It happens often enough for the time to add up. - Documented. You've written the steps down, as above. In most online stores, the best first candidates are order tagging and routing, low-stock alerts, review requests after delivery, abandoned checkout emails and a weekly sales summary. Leave refund disputes, supplier negotiations and anything that needs judgement with a person for now. ### Check what you already have for free Before you pay for anything, look at what your platform includes. On Shopify, Flow is a free app on the Basic, Grow, Advanced and Plus plans (Shopify (https://help.shopify.com/en/manual/shopify-flow)), and it can handle a lot of the tagging, alerts and routing above at no cost. Its Send HTTP request action, which lets it talk to other apps, only comes with Grow and above. A free automation still costs setup time, but the payback maths gets very easy when the build cost is close to zero. ### What to measure for each automation Decide the measure before you build, and write down today's number. Otherwise, three months later, nobody can say whether it worked. AutomationMeasure before and after Order processingMinutes per order, orders with mistakes Low-stock alertsDays out of stock on your top sellers Abandoned checkout emailsRevenue attributed to the flow Review requestsReviews collected per 100 orders Weekly reportingHours spent putting the report together ### When automation isn't worth it Sometimes the honest answer is no. Skip it, or wait, when: - The task happens a few times a month. The setup will never pay back. - The process is still changing. Automate it once it's settled. - The real problem is upstream. Automating a messy returns process gives you a faster messy returns process. - It needs judgement on most cases. Keep a person on it and automate the paperwork around them. ### Where to start Write down your three most repetitive tasks, time each one for a week, and run the calculation above on all three. The one with the shortest payback is your first automation. If you'd rather have someone go through the store with you, our Store Ops Audit looks at your current automations, apps and processes and ranks what to fix first by money against effort. ### AI Chatbot vs Live Chat: When to Use Each https://www.axioprax.com/insights/ai-chatbot-vs-live-chat (2026-08-29) AI chatbots and live chat solve different problems, and the wrong choice costs you in service or overhead. How to decide which one fits your business. Both promise the same thing: customers get answers without waiting for an email reply. They're not interchangeable, though. An AI chatbot is good at answering the same questions all day. A person on live chat is good at everything that needs judgement. Pick the wrong one for your mix of questions and you either frustrate customers or pay for people to type the same answer fifty times a day. Here's how to decide, starting with the questions you actually get. ### Start with your questions, not the tool Pull a month of chats or support emails and sort them into two piles. The first pile is predictable: order status, delivery times, returns, opening hours, pricing, how to do something in your product. The second needs judgement: a complaint, a product comparison with unusual requirements, a custom order, a refund outside the rules. If most of your volume is in the first pile, a chatbot will take real work off your team. If most is in the second, a chatbot mostly gets in the way, and live chat staffed by someone who knows the product is the better spend. ### What each one is good at An AI chatbot answers at any hour, gives the same answer every time, and costs roughly the same whether it handles ten conversations or a thousand. It's only as good as what you've written down for it, and it should hand over the moment it's out of its depth. Live chat with a person handles the conversations that decide whether someone buys or comes back: the hesitant buyer, the unhappy customer, the unusual request. It costs staff time, and it only works during the hours someone is there to answer. ### What each one costs Live chat costs people's time. As a rough guide, the US Bureau of Labor Statistics put the median wage for customer service representatives at $21.53 an hour in May 2025 (BLS (https://www.bls.gov/ooh/office-and-administrative-support/customer-service-representatives.htm)). Covering chat for a 40-hour week at that rate is about $3,700 a month in wages alone, before benefits, management or evening cover. A chatbot costs a build or a subscription, plus usage that grows with the number and length of conversations. The bigger cost people forget is upkeep: someone has to keep its knowledge current and read a sample of its conversations every week. Budget for that time, or the bot slowly drifts out of date. ### Usually the answer is both For most businesses the right setup is a chatbot first and a person behind it. The bot answers the predictable questions. Anything it can't answer, or anything where the customer is upset, goes to a person with the conversation attached, so nobody has to repeat themselves. Two details make this work. First, the bot should hand over early rather than late; three failed attempts to answer is two too many. Second, outside your opening hours it should say plainly when a person will reply and take the details, instead of pretending someone is about to join. ### By type of business - Online store. Chatbot for order status, returns and delivery questions. A person for complaints and damaged or wrong items. - Service business, such as a clinic, salon or consultancy. Chatbot for booking, prices and availability. A person for anything clinical, legal or sensitive. - B2B sales. A person for the sales conversation itself. A chatbot can answer product questions after the sale. - Software product. A chatbot that answers from your documentation. A person for account problems and bugs. ### How to tell if it's working Whichever you choose, measure three things for the first two months: how long customers wait for a first reply, how many conversations end with the question answered, and how many complaints mention the chat itself. For a chatbot, also read a sample of transcripts every week. Most bad answers trace back to something missing from its knowledge, which our guide to training a chatbot on your business (/insights/train-ai-chatbot-business-knowledge) covers. For Shopify stores, our Shopify chatbot guide (/insights/shopify-ai-chatbot-guide) shows the options and how to answer order questions safely. ### How to Train an AI Chatbot on Your Business https://www.axioprax.com/insights/train-ai-chatbot-business-knowledge (2026-08-29) A generic chatbot is not useful. How to build the knowledge base that lets an AI chatbot answer your customers’ real questions accurately from day one. When a business chatbot gives a wrong answer, the model is rarely the problem. The problem is what it was given to work from: a thin FAQ, a returns policy from two years ago, and nothing about the questions customers actually ask. The bot either guesses or says "I don't know" to something it should have known. "Training" a chatbot for a business almost never means retraining the AI model. It means giving it a well-organised set of your own information to look things up in before it answers. Here's how to build that, test it and keep it current. ### What "training" actually means Most business chatbots work by retrieval. When a customer asks something, the system searches your knowledge for the most relevant passages and gives them to the AI model along with the question. The model writes the answer from those passages. If the right passage isn't there, the answer can't be right either. Think of it as the briefing pack for a new team member who starts answering customers on their first morning. The better the pack, the better the answers. ### Where the knowledge comes from - Your policies. Returns, shipping, refunds, warranty and privacy, in plain language. - Product or service details. Specifications, sizes, compatibility, ingredients, what's included in each package. - Your FAQ, with every vague answer expanded into full sentences. - Past support conversations. The best source of the questions people really ask, in the words they really use. - Your team's standard replies. The answers people already type from memory. Start with the last three months of support emails or chats. Group the questions by topic and count them. A small number of topics usually covers most of the volume. Make sure each one has a complete, current answer before you worry about the long tail. ### Write it so it can be found The way you write the knowledge matters as much as what's in it. Keep each topic in its own section with a clear heading. Put the answer in the first sentence and the conditions after it. Use the words customers use: if they say "send it back", don't only write "return merchandise authorisation". Three things cause most wrong answers: - Vague policies. "Returns on most items" invites the bot to fill in the gaps. - Missing edge cases. Damaged on arrival, wrong size, missed the return window. These generate a lot of messages, so write them down. - Old versions. If last year's policy is still in there next to this year's, the bot may quote either. Delete old versions, and put a "last updated" date on each policy. ### Test it with real questions before launch Take fifty real questions from your support history, including awkward ones and a few that your knowledge deliberately doesn't cover. Run each through the bot and mark the answer: right, wrong, or handed over to a person. The questions it shouldn't be able to answer matter most. A good bot says it doesn't know and offers a person. A bad one makes up something plausible. If yours invents answers, tighten its instructions so it answers only from your knowledge, and run the set again. ### Keep it current after launch For the first month, read a sample of conversations every week. List the questions it got wrong or couldn't answer, fix the knowledge behind each one, and check the fix by asking the same question again. After that, a monthly review is usually enough. Give the knowledge an owner. When a policy changes, a product launches or prices move, updating the bot should be on the same checklist as updating the website. Most bots that "got worse" simply fell out of date. If you're still deciding whether a bot is right for you, see AI chatbot vs live chat (/insights/ai-chatbot-vs-live-chat). For a store, our Shopify chatbot guide (/insights/shopify-ai-chatbot-guide) covers order questions and the options available today, and the same knowledge principles apply on WhatsApp, as our WhatsApp automation guide (/insights/whatsapp-business-automation-guide) explains. ### WhatsApp Business Automation: A Practical Guide for Shopify Stores and SMBs https://www.axioprax.com/insights/whatsapp-business-automation-guide (2026-08-30) What WhatsApp automation can do, the opt-in and template rules, what changed in Meta pricing on 1 October 2026, and a worked Shopify order update flow. If your customers live on WhatsApp, an order update there gets seen. The same tracking link in an email is competing with everything else in their inbox. That's most of the appeal. The catch is that WhatsApp plays by stricter rules than email, and Meta changed what it charges on 1 October 2026. A setup that cost almost nothing in September costs more now. This guide covers what you can automate, the rules you can't skip, what the October change means in money, and how we'd set up order updates for a Shopify store. Everything about Meta's rules and prices below comes from Meta's own pages, read on 20 October 2026. ### The app or the API? There are two WhatsApps for business, and only one of them can really be automated. The WhatsApp Business app, the one on your phone, has quick replies, a greeting and an away message. It's handy, but it's made for a person typing. Anything that sends messages from your store or your other systems needs the WhatsApp Business Platform, which is Meta's API. The rest of this guide is about the API. ### What businesses use it for - Order confirmations, shipping notices and delivery updates - Booking confirmations and reminders - An AI assistant for the common questions, with a person behind it - Payment reminders and receipts - Marketing messages, to people who asked for them - A daily summary for the owner ### Three rules you can't skip Opt-in. You can only message someone who gave you their number and agreed to hear from you on WhatsApp. That's in the WhatsApp Business Messaging Policy (https://business.whatsapp.com/policy), and a box ticked for email marketing at checkout doesn't cover it. Ask for WhatsApp on its own and keep a record of the answer. Shopify can now do this for you: since September 2026 it can collect WhatsApp marketing consent at checkout, as a separate box (Shopify Changelog (https://changelog.shopify.com/), 10 September 2026). Templates. If you start the conversation, the message has to be a template Meta has approved. There are three kinds: marketing, utility (an order shipped, a booking confirmed) and authentication (login codes). The 24-hour window. When a customer messages or calls you, a 24-hour window opens, and every new message from them resets the clock. Inside it you can reply in your own words. Once it closes, you're back to templates. ### What changed on 1 October 2026 Meta has charged per delivered message since July 2025. The rate depends on the kind of message and on the country of the person receiving it. Then on 1 October 2026, two things that used to be free stopped being free (Meta's pricing documentation (https://developers.facebook.com/documentation/business-messaging/whatsapp/pricing)): MessageUntil 30 September 2026From 1 October 2026 Utility template inside the 24-hour windowFreeCharged at the utility rate Service message (your own reply, typed by a person or written by a third-party AI)Free1,000 free a month per business number, then charged. Unused ones don't roll over. Marketing and authentication templatesChargedCharged One free route is left. If someone taps a Click to WhatsApp ad in the mobile app and you reply inside the window, that conversation is free for up to 7 days. In plain terms, if your store sends a WhatsApp message for every tracking scan, you now pay for every one of them. That's not a reason to stop. It's a reason to be choosy. Meta lists the rate for each country on the same page, and a solution provider, if you use one, may add its own fee on top. ### What you need before you start - A Meta Business account - A business phone number registered on the platform - Access, either directly through Meta's Cloud API or through a Business Solution Provider - An approved template for every message you'll send first - Somewhere to record each customer's WhatsApp opt-in ### Can Shopify send WhatsApp messages itself now? For marketing, yes. Since 29 July 2026, WhatsApp marketing is part of Shopify Messaging, the same app that sends Shopify's marketing emails. You can send campaigns from pre-built templates or your own content, add products from your catalog and buttons, and reply automatically when someone taps a button or sends a keyword. Shopify says you pay only for the messages you send (Shopify Changelog, 29 July 2026 (https://changelog.shopify.com/posts/shopify-messaging-now-supports-whatsapp-marketing)). On 28 October 2026 Shopify added WhatsApp marketing automations to Shopify Messaging as well (Shopify Changelog (https://changelog.shopify.com/)). Shopify's help page describes these as marketing messages to subscribers who opted in to WhatsApp marketing from your store (Shopify Help Center, WhatsApp in Shopify Messaging (https://help.shopify.com/en/manual/promoting-marketing/create-marketing/shopify-messaging/whatsapp)). It does not describe order updates, support conversations or an AI assistant. So if you only want WhatsApp campaigns, start with Shopify Messaging. For order updates and replies to customer questions, you still need an app or a workflow like the one below. ### Order updates for a Shopify store This is where we'd start almost any store, because it answers the question customers would otherwise email you about: where's my order? Shopify keeps a record of each step a shipment goes through, called fulfilment events. The statuses include IN_TRANSIT, OUT_FOR_DELIVERY, DELIVERED, DELAYED and FAILURE (Shopify's developer docs (https://shopify.dev/docs/api/admin-graphql/latest/enums/FulfillmentEventStatus)). A workflow can watch for those, and each time one comes in: - Check the customer opted in to WhatsApp. If they didn't, stop there and let email do the job. - Pick the approved utility template that matches the status. - Send it through the Cloud API or your provider. - Leave a note on the order saying what went out and when. Now that utility messages cost money, we wouldn't send all of them. Our shortlist: - Shipped, with the tracking link. People expect this one. - Delayed or failed. This is the message that saves you a support ticket. - Delivered. And only ask for a review if the order actually arrived on time. Asking someone whose parcel was late to rate you is a quick way to collect one-star reviews. You can build this with a Shopify app or with a workflow tool like n8n. If you're weighing those up, our n8n vs Zapier (/insights/n8n-vs-zapier-honest-comparison) piece runs one store order flow through both. It works the other way round too. A short message to the owner each morning, with yesterday's sales against a normal day, low stock and anything shipped late, is a lot harder to ignore than an emailed report. ### AI replies on WhatsApp An AI assistant can take the questions that come in all day inside the 24-hour window: delivery times, returns, sizes, where's my parcel. Under the October rules its replies count as service messages, so the first 1,000 a month per number are free and after that you pay. Two things we'd insist on. It answers only from your own policies and order data, and it hands over to a person cleanly when it's out of its depth. Building an AI chatbot for a Shopify store (/insights/shopify-ai-chatbot-guide) goes into the data side. ### If you only do one thing Set up the delayed or failed delivery message. One template, it only fires when something has gone wrong, and it replaces an email you'd otherwise write by hand. Once that's running, add shipped and delivered. ### Accounts Receivable Automation: Stop Chasing Late Invoices https://www.axioprax.com/insights/accounts-receivable-automation-guide (2026-08-30) A reminder schedule that runs itself, what QuickBooks, Xero and Stripe already do for free, what automation adds on top, and where a person still needs to step in. Chasing late payments is one of those jobs that never shows up on a job description but eats a surprising amount of time. In the UK, government research found that business owners affected by late payment spend 86 hours a year chasing invoices (GOV.UK, March 2026 (https://www.gov.uk/government/consultations/late-payments-tackling-poor-payment-practices/outcome/late-payment-consultation-time-to-pay-up-government-response-web-version)). That's more than two working weeks a year spent writing polite reminder emails. The good news is that most of that chasing follows the same pattern every time, which means it can be automated. This guide explains how, what your accounting software already does for free, and where a person still needs to step in. ### Where the time actually goes The reminder email itself takes a minute. The time goes on everything around it: checking who's overdue, finding the invoice, writing something that's firm but not rude, logging that you sent it, and remembering to try again next week. Multiply that by every late customer and every month, and the 86 hours start to make sense. ### A reminder schedule that works The schedule matters more than the wording. This is the one we'd start with, and adjust once you see how your customers respond: - 3 days before the due date. A friendly heads-up with the amount and a payment link. Many late payments are just forgotten ones. - On the due date. Short and neutral: today's the day. - 7 days late. Still polite, a little firmer, with the invoice attached again. - 14 days late. Clear about what happens next, whether that's paused work, a late fee or a phone call. - 30 days late. The automation stops and a person picks up the phone. Two rules make this work. Reminders must stop the instant the invoice is paid, because nothing annoys a customer more than a reminder for a bill they've settled. And they must pause when a customer raises a dispute, so a person can deal with it. ### What your accounting software already does Before you build anything, switch on what you already pay for. - QuickBooks Online can send up to three automatic reminders, each with its own wording, scheduled anywhere up to 90 days before or after the due date. They only go to invoices you've already emailed that are still unpaid (QuickBooks (https://quickbooks.intuit.com/learn-support/en-us/help-article/invoicing/send-invoice-reminders-automatically-manually/L84cQjpxo_US_en_US)). - Xero sends reminders automatically before or after the due date, on a schedule and with a message you set, and stops as soon as payment arrives (Xero (https://www.xero.com/us/accounting-software/invoice-reminders/)). - Stripe Invoicing charges 0.4% per paid invoice on Starter or 0.5% on Plus, on top of card fees, and includes a hosted payment page, automatic reminders and smart retries for failed payments (Stripe (https://stripe.com/invoicing/pricing)). For a lot of small businesses, turning these on is most of the job. It costs nothing and you can do it this afternoon. ### What automation adds on top The built-in reminders cover the emails. A proper AR workflow adds the parts around them that still take time: - A payment link in every reminder, so paying takes one click. - Dispute handling that pauses reminders and tells the right person. - A heads-up for big invoices. A message in Slack or by email when a large invoice goes overdue, so you hear about it before it becomes a problem. - Notes in your CRM, so whoever talks to the customer next knows they owe money. - A short daily summary of who owes what and how late, instead of opening the aged receivables report. ### Where a person still matters Automation is for the routine cases. Keep a person on disputes, payment plans and your most important accounts. A long-standing customer who's suddenly 30 days late deserves a phone call, not a fourth automated email. If anything, automation should make those calls easier by flagging them sooner. ### If you trade in the UK You can already charge statutory interest on late business payments at 8% above the Bank of England base rate, plus fixed compensation of £40, £70 or £100 depending on the size of the debt (GOV.UK (https://www.gov.uk/late-commercial-payments-interest-debt-recovery/charging-interest-commercial-debt)). The government has also announced a 60-day maximum payment term, starting no earlier than 2027, and new powers for the Small Business Commissioner to fine persistent late payers. These are announced but not yet law, so check before you rely on them. ### Where to start This week, switch on the reminders in your accounting software and set the schedule above. Watch it for a month. If you're still chasing by hand after that, the remaining work, such as payment links, dispute handling and alerts, is a short, well-defined automation to build next. ### How to Automate Monthly Financial Reporting (and Your Supplier Bills) https://www.axioprax.com/insights/financial-reporting-automation (2026-08-30) How a scheduled workflow builds your monthly report from Xero or QuickBooks, your CRM, Shopify and the bank, what changed in their API pricing, and how to automate supplier bills too. On the first working day of the month, someone in most small businesses opens the accounting software, the CRM, the bank and maybe Shopify, copies numbers into a spreadsheet, tidies it up and emails it round. It takes a few hours, it looks the same every month, and it's exactly the kind of job a computer should be doing. This guide covers how monthly reporting automation works, the one thing to settle before you build it, what changed in Xero's and Intuit's API pricing over the past year, how to handle revenue that comes from Shopify, and how to automate supplier bills while you're at it. Facts were checked on the vendors' own pages on 20 October 2026. ### What goes in the report A good monthly report for a small business fits on one page. Usually it covers: - Revenue against budget and against last month - Gross margin and cost of goods sold - Operating costs by category - Cash in the bank and how long it lasts - Who owes you money, and how late they are - What you owe, and when it's due - Two or three business numbers you actually steer by, such as orders, pipeline or churn ### How the automation works The workflow runs on a schedule, usually early on the first working day of the month, and does four things. - Pull the data. Xero's API has ready-made report endpoints for profit and loss, balance sheet, and aged receivables and payables (Xero (https://developer.xero.com/documentation/api/accounting/reports)). QuickBooks Online has a Reports API that covers the same ground. Pipeline numbers come from your CRM, and store revenue from Shopify. - Calculate. Variances against budget and last month, margins, and how many days of cash you have. - Format. The same template every month, as a PDF, a Google Doc or a plain email. - Send and file. To the owner, the finance lead or the board, with a copy saved to a shared folder. ### Settle what every number means first This is where reporting automation usually goes wrong, and it has nothing to do with technology. "Revenue" can mean gross sales, sales after refunds and discounts, revenue recognised in the period, or cash that actually arrived. All four are reasonable. They give different answers for the same month. If the report calculates a number differently from how the owner reads it, the automation creates confusion instead of saving time. So before anything gets built, write a one-line definition for every number on the report and get the person reading it to agree. Then lock the format. An automated report expects the same layout every month. ### What changed in Xero and QuickBooks API pricing Both big accounting platforms changed how they charge apps for API access over the past year. It's worth knowing before you build. - Xero moved to new developer tiers on 2 March 2026 (Xero FAQ (https://developer.xero.com/faq/pricing-and-policy-updates)). Starter is free for up to 5 connections. Paid tiers start at A$35 a month for 50 connections (Xero pricing (https://developer.xero.com/pricing)). Integrations built for your own business or a single client are exempt, and so are custom connections. Xero's updated terms also stop developers from using data from its APIs to train AI models. - Intuit launched its App Partner Program on 28 July 2025 (Business Wire (https://www.businesswire.com/news/home/20250728896195/en)). Sending data into QuickBooks is free, but reading data out, including fetching reports, is metered. The free Builder tier includes 500,000 of those calls a month. Variable fees started in the US on 1 November 2025, and the programme reached the UK, Australia and Canada on 3 November 2025 (Intuit (https://blogs.a.intuit.com/?p=35224)). For a business automating its own monthly report, neither change is likely to cost anything. They matter more if you're an agency or app connecting many clients. ### If your revenue comes from Shopify Store owners often report Shopify payouts as revenue. They're not the same thing. A payout is what's left after Shopify takes fees, refunds and chargebacks, sometimes across several days of orders. Shopify's API exposes each balance transaction with its amount, fee, net figure, type and the payout and order it belongs to (Shopify (https://shopify.dev/docs/api/admin-graphql/latest/objects/ShopifyPaymentsBalanceTransaction)). That's enough for an automation to break every payout into sales, refunds, fees and adjustments, match it to the bank, and show true revenue and fees in the monthly report. Shopify's own Finances reports in the admin are a good place to check the totals against. ### Automate the bills while you're at it The other half of the month-end grind is supplier bills: downloading invoices, typing them in, checking them against what was ordered, getting them approved and paying them. Automation handles this in five stages. - Capture. Both platforms now do part of this out of the box. QuickBooks lets you forward bills to a dedicated email address and pulls out the date, amount and supplier (QuickBooks (https://quickbooks.intuit.com/learn-support/en-us/help-article/accounts-payable/email-receipts-bills-quickbooks-online/L7r2LAQ7C_US_en_US)). Xero's smart document capture is included on all its small business plans and takes bills by email, photo or upload (Xero (https://www.xero.com/uk/accounting-software/capture-data-with-hubdoc/)). - Extract. AI reads the line items, totals, tax and due date. Be wary of accuracy claims. We couldn't find independent data on how accurate invoice extraction is, only vendors' own figures, so keep a person checking anything from a new supplier or an unusual layout. - Match. Check each bill against the purchase order. Skipping this is easier to build and a real fraud risk. - Approve. Small, matched bills go straight through. Large ones, or anything from a new supplier, wait for a named person. - Post and pay. The approved bill goes into the accounting software and the payment gets scheduled. One thing people forget: alerting. If the inbox connection drops, bills stop flowing silently. Build an alert from day one. ### A rule change worth knowing about If you invoice in the UK, the government has confirmed mandatory e-invoicing for all VAT invoices from 2029, with a detailed roadmap promised at Budget 2026 (GOV.UK (https://www.gov.uk/government/consultations/promoting-electronic-invoicing-across-uk-businesses-and-the-public-sector)). In the EU, digital reporting for cross-border business sales starts on 1 July 2030 (European Commission (https://taxation-customs.ec.europa.eu/news/adoption-vat-digital-age-package-2025-03-11_en)). Neither is urgent today, but automating your bills and invoices now means you won't be scrambling later. ### Where to start Write down the report you want, line by line, with a definition for each number. List where each number lives. Then automate the pull for one month and compare it with the report you made by hand. When the two match, switch off the manual version. ### Building AI Agents That Work in Production https://www.axioprax.com/insights/designing-ai-agents-that-work-in-production (2026-08-24) AI agents fail in production for different reasons than in demos. Task decomposition, tool design and error handling for agents that work reliably. An AI agent that impresses in a demo and one that runs your operations unattended are different things. The demo shows the happy path. Production is everything else: the input in the wrong format, the API that times out, the task that can't be finished because something upstream changed. Most of what makes an agent reliable isn't clever prompting. It's a handful of engineering decisions made before the first run. Here they are, starting with the one most teams skip: whether you need an agent at all. ### Start with a workflow, not an agent Anthropic draws a useful line. In a workflow, the model and its tools follow paths you've written in code. In an agent, the model decides its own steps and which tools to use. Their advice is to find the simplest solution that works and add complexity only when it's needed, because agents trade cost and speed for flexibility, and their errors can compound (Anthropic (https://www.anthropic.com/research/building-effective-agents)). In practice, most business tasks are workflows. Processing an invoice follows the same steps every time, and the model's job is to read and judge within each step. Research that needs to follow leads across sources, where the next step depends on what the last one found, is where an agent earns its place. ### Break the task into steps you can check Give an agent one broad instruction and it makes dozens of small decisions about scope and format, differently on every run. Give it a sequence of defined steps, each with an expected input, an expected output and a clear failure condition, and you get consistent results. It also makes problems fixable. "Which step went wrong?" has an answer. "Why did the agent do the wrong thing?" usually doesn't. ### Design tools the model can't misuse Anthropic recommends putting as much care into the interface between the model and its tools as you'd put into a screen for people: clear names, descriptions written like documentation for a new colleague, and arguments shaped so mistakes are hard to make. The pattern we use is a thin adapter around every external system. The model never calls the raw API. It calls an adapter that checks the input, makes the call, turns the response into a small, predictable shape and returns errors as plain results the model can read, instead of crashing. ### Decide where a person approves Whether an agent needs approval for an action is a decision to make before launch, based on two questions: how much does it cost if this is wrong, and can it be undone? Drafting an email or updating a note can run on its own. Sending to a customer, issuing a refund, paying a bill or deleting a record should wait for a person. Holding an order sits in between: it's easy to undo, so it can run and tell someone. That's the pattern in our Fraud Hold demo build (/insights/fraud-hold-demo-build), where a rule holds a risky order and only a person releases it. ### Put limits on every run An agent that gets the right answer after forty tool calls when five would do is still a problem. Costs and waiting times add up. Set a maximum number of steps and a cost ceiling per run, and when the agent hits either, it should stop and say so rather than keep going. ### Make failure obvious The most dangerous agent is one that fills a gap with something plausible. When it doesn't have what it needs, it should return a clear "couldn't complete" with the reason, not a guess that looks finished. Log every tool call with its inputs and result. Before launch, build a test set from real past cases, including the awkward ones, and run it after every change to the prompt or tools. An agent is ready when it passes that set reliably and fails loudly on the cases it can't handle. ### Where to go next Agents usually connect to other systems through a standard such as MCP, which our guide to MCP as an integration layer (/insights/mcp-the-integration-layer-ai-systems-need) explains. For the code that holds an agent together, prototype to production (/insights/prototype-to-production-ai-generated-code) covers what to check before it goes live. ### MCP: The Integration Layer Every AI Agent Needs https://www.axioprax.com/insights/mcp-the-integration-layer-ai-systems-need (2026-08-25) MCP standardises how AI agents connect to external tools. What it solves, where it fits in an agent architecture and practical patterns for production builds. An AI assistant that can read your calendar, look up a customer and update a record has to connect to three systems, each with its own login, data format and quirks. For years, every one of those connections was built by hand, separately for each AI product. The Model Context Protocol, or MCP, is the open standard that changed that. Here's what it is, who supports it now, where it fits in a real system, and the security questions to ask before you connect anything to it. ### What MCP is Anthropic introduced MCP in November 2024 as an open protocol for connecting AI applications to tools and data. In December 2025 it was donated to the Agentic AI Foundation, a fund under the Linux Foundation, co-founded by Anthropic, Block and OpenAI. At that point MCP was supported in ChatGPT, Claude, Cursor, Gemini, Microsoft Copilot and Visual Studio Code, with more than 10,000 active servers (MCP blog (https://blog.modelcontextprotocol.io/posts/2025-12-09-mcp-joins-agentic-ai-foundation/)). So it's no longer one company's proposal. It's the common way AI tools connect to other systems. ### What it solves Without a standard, every AI app needs its own connector to every system. Five AI tools and ten business systems means up to fifty separate integrations, each maintained on its own. With MCP, each system is wrapped once in an MCP server, and any AI app that speaks MCP can use it. ### How it works An AI application, the client, talks to MCP servers. Each server sits in front of one system and offers three kinds of thing: - Tools are actions the AI can take: search orders, create a ticket, send a message. - Resources are data it can read: documents, records, files. - Prompts are ready-made instructions for common jobs with that system. The server handles logging in, calling the API and shaping the response. If the system's API changes, you update the server, and every AI app using it keeps working. You can see this in commerce already. Shopify publishes Storefront MCP, which lets an AI assistant search a store's catalogue, manage a cart and answer policy questions, with a separate server for customer orders and returns (Shopify.dev (https://shopify.dev/docs/apps/build/storefront-mcp/index)). Our Shopify chatbot guide (/insights/shopify-ai-chatbot-guide) shows where that fits. ### When MCP is worth it, and when it isn't MCP pays off when several AI tools need the same systems, when you want to change AI providers without rebuilding connections, or when a vendor already publishes a server you can use. For a single job that calls one API in one way, it's often simpler to call that API directly from your workflow. An n8n workflow that posts a Slack message when an order fails doesn't need MCP. Pick MCP when the connection will be reused, not because it's new. ### Security comes first Connecting an AI to your systems gives it the power to act in them, so treat every server like a new employee with keys. The OWASP MCP security cheat sheet lists the main risks: instructions hidden in tool descriptions or tool results, servers that change their tools after you approved them, permissions far broader than the job needs, and data leaking out through ordinary-looking tool calls (OWASP (https://cheatsheetseries.owasp.org/cheatsheets/MCP_Security_Cheat_Sheet.html)). The practical rules: install servers only from sources you trust, give each one the narrowest access it needs, review its tools again when they change, treat anything a tool returns as untrusted, require a person's approval for sending, paying, refunding or deleting, and log every call. Most important, enforce permissions in your own code. Don't rely on the AI to police itself. ### Where to start List the systems your team would want an AI assistant to use, and check which already have an official MCP server. Start with read-only access to one of them, watch the logs for a few weeks, and add actions only once you trust what you see. For the wider design of systems that act on their own, see building AI agents that work in production (/insights/designing-ai-agents-that-work-in-production). ### Prototype to Production: What AI Code Gets Wrong https://www.axioprax.com/insights/prototype-to-production-ai-generated-code (2026-08-26) AI code generation is fast but often fragile. Review patterns for error handling, security and edge cases that turn AI prototypes into production code. AI coding tools have changed how fast a prototype appears. Something that used to take a week can be on screen in a day, built by someone who isn't a senior engineer. That's genuinely useful. The catch shows up later: code that works for ten test users can hide problems that only appear with real users, real data and real attackers. This guide covers what AI-generated code tends to get right, where it goes wrong, what the research says about security, and a review order that catches the expensive problems without rereading every line. ### What it gets right, and where to look harder AI tools are good at well-trodden patterns: standard API endpoints, ordinary database queries, interface components in popular frameworks, config files, and turning one format into another. For this kind of code the first draft is often close to finished. The problems sit at the edges, wherever the real situation differs from the common pattern the tool has seen many times. Error handling. The happy path works. Failures are often caught and logged with a vague message, then ignored, when the right response is to retry, alert someone or stop. Silent failures are the hardest bugs to find in production. Edge cases. Code written from a short description handles the case in the description. Empty inputs, missing values, odd characters and very large numbers rarely make it in, and those are exactly what real users send. Security. Missing input checks, queries built by gluing strings together, overly open access settings and sensitive data written to logs all appear regularly. Scale. Patterns that are fine for a demo, such as a database query inside a loop or a missing index, can slow to a crawl with real volumes. ### What the research says about security Veracode tested more than 100 large language models on 80 coding tasks. About 45% of the generated code samples contained security flaws, many from the OWASP Top 10. Java fared worst, failing more than 70% of the time, with Python, C# and JavaScript between 38% and 45%. The models failed to defend against cross-site scripting in 86% of relevant cases (Veracode, July 2025 (https://www.veracode.com/press-release/ai-generated-code-poses-major-security-risks-in-nearly-half-of-all-development-tasks-veracode-research-reveals/)). That's one vendor's test set, and models change quickly, so treat the exact figures as a snapshot. The direction is the point: AI-written code needs a security review before it handles real users. ### The review order we use Rereading every line isn't practical and isn't where the risk is. Review in this order, and stop to fix before moving on: - Who can call what. Every endpoint and action checks who is asking and whether they're allowed. - Inputs from outside. Anything from users, files or other APIs is validated before use, and database queries are parameterised. - Failure paths. For every external call: what happens when it fails, times out or returns nonsense? - Secrets and personal data. No keys in the code, nothing sensitive in logs or in responses that don't need it. - Data access at scale. Queries, indexes and anything inside a loop. - Tests for the edge cases. Written down and run before go-live, not just tried once by hand. Automated checks help with the routine parts: a linter, a dependency scanner and a secret scanner in your build will catch a lot before a person looks. They don't replace the first three steps. ### Treat it like a capable new colleague The most useful mental model is a fast, capable junior developer. You'd let them draft quickly and you wouldn't ship their work unreviewed. Use AI for the first draft and the repetitive parts, then review with the same care you'd give any pull request, focused on the places above. For systems where AI does more than write code and starts taking actions, building AI agents that work in production (/insights/designing-ai-agents-that-work-in-production) covers the guardrails, and our MCP guide (/insights/mcp-the-integration-layer-ai-systems-need) covers connecting AI to other systems safely.