No Shopify Store in 122 Passed All 9 AI-Agent Checks

Search Engine Journal tested 122 top Shopify stores for AI-agent readiness, and none passed all nine checks.

If you run a high-ticket store, the failures landed exactly where you make money. The shipping question, the cancellation question and the open-box return question all came back empty or unusable from nearly every store. This is the first public audit I have seen that measures what an AI shopping agent actually hears when it asks your store a question, and I run this site, Ecommerce Paradise, for people who sell $1,500 to $15,000 products.

The audit, published October 14, 2026 by Bill Hunt, President at Bisan Digital, found that Shopify installs the plumbing by default and merchants are not filling it with real answers. Below are the numbers, the context, the one fair counterpoint, and the six fixes I would make this week. If you are still deciding what to sell, start with my guide to what high-ticket dropshipping is, because the stakes of this audit rise with order value.

Fix your store’s front end now, but fix the legal entity behind it too: form your LLC with Northwest Registered Agent, the best LLC formation service for ecommerce owners. Northwest has run for roughly 25 years under stable ownership, which beats a formation mill that changes hands and renewal prices. Form your LLC with Northwest →

122 Shopify Stores Tested: Zero Passed All 9 Agent Checks

Hunt’s sample was 122 leading Shopify storefronts, drawn from published rankings of top Shopify merchants and probed from August to September 2026, according to his Search Engine Journal write-up. He tested agents.md, llms.txt, UCP profiles and MCP tool listings, product-page structured data, policy text, questions put to each store’s policy tool, a non-completing checkout, and a static-HTML accessibility check.

Coverage varied by test. Policy text was readable for 56 stores, the policy tool for 88, live checkout for 74 and product-page accessibility for 68. Hunt records throttled or blocked responses as unknown, not absent.

The platform layer worked. Per the audit, 76 of 86 product pages loaded Shopify’s WebMCP adapter script, and 89 of 122 stores served an agents.md file. All 74 stores that were checked created a checkout, returned typed statuses, gave coded errors and handled cancellation correctly. Every one returned requires_escalation, which means a human approves the final purchase.

The merchant layer did not. Of those 89 agents.md files, 83 were near-identical copies of one Shopify template, and only five were written by merchants. None of the 122 stores passed all nine decision-readiness checks, and the best three reached seven.

What agents heard when they asked policy questions

Hunt put four questions to each store’s policy tool. Return window: 88 stores returned something and 70 gave a usable answer. Shipping cost and arrival to a Boston address: one store returned something and zero were usable. Cancellation: 8 returned something and 2 were usable. Used or open-box returns: 88 returned something and 5 were usable.

Delivery data was thin even at checkout. Shipping was priced for 36 of 74 stores, and none returned an arrival estimate or carrier. All 93 stores that declared UCP fulfillment listed shipping only, with no pickup option.

What agents found on the product page

Among the 68 stores in the accessibility check, 47% had at least one unlabelled input, 46% had an empty link and 37% had an empty button. Of stores using dropdowns, 65% had an unlabeled size or option selector. Nineteen of the 68 had no add-to-cart control in server HTML, and 31 of 68 passed the full floor.

Structured data told the same story. According to the audit, 70% of stores with readable policy text stated return windows in prose, but only 7% exposed merchantReturnDays. And none of the 122 stores offered a tool to cancel, change or return an order. Another 51 of 122 blocked /policies/ in robots.txt.

How Shopify Built the Agent Layer Merchants Now Have to Fill

Shopify started generating agents.md and the WebMCP adapter script for Liquid storefronts on August 5, per Hunt’s audit. That explains the template copies: the file exists on most stores because Shopify put it there, not because anyone wrote it.

On September 28, Shopify extended WebMCP into checkout. According to the Shopify developer changelog, browser agents can now read and update the checkout in a shopper’s tab through four tools: get_checkout, update_checkout, complete_checkout and a return-to-storefront tool. Shopify says the tools “don’t expose a new API or require merchant configuration.”

The limits matter. Per Search Engine Journal’s report on that launch, agents cannot add new card details or change order items, and 3D Secure or login challenges hand control back to the buyer. Standard three-page checkout gets no tools unless the buyer uses Shop Pay. B2B, draft order and order edit checkouts are excluded.

Shopify’s own test, as summarized there, ran ten checkout tasks across two test shops. WebMCP succeeded in 60 of 60 attempts versus 56 of 60 for browser automation, at 10.3 seconds versus 27.4 seconds on average. The test used one model and contains no real-world order or conversion data.

This is the latest step in a thread I have been tracking. I covered Shopify letting AI agents submit orders on your store. Then came Shopify opening its catalog to personal AI agents.

I also flagged the security hole in the AI running Shop Pay checkout. Shopify launched a free readiness tool in April that runs 31 checks across five categories, per Shopifreaks.

The fair counterpoint

Hunt lists his own limits. The September live test used Shopify’s server route, not the browser route another tester used in August, so it does not show that an August defect was fixed. Some failures came from stale item IDs collected weeks earlier. The browser WebMCP test covered two stores and submitted no orders.

The largest caveat is that the test handed agents exact item IDs. Real buyers need help choosing products, and this audit did not measure that. Shopify’s position, in its changelog, is that merchants need to configure nothing for the checkout tools to appear. Both statements can be true: the tools appear, and the answers behind them are empty.

Why High-Ticket Stores Lose When Agents Cannot Quote Freight

My read: the zero usable shipping answers is the number that matters most, and it matters more for us than for a $30 product store. A cheap item ships in a parcel with a flat rate. A $3,000 patio set or a $6,000 mobility scooter ships freight, often on a pallet, with lift-gate fees and a delivery window the buyer cares about. A shopper asking an agent “how much to ship to Boston and when does it arrive” is asking the exact question that decides a high-ticket purchase.

When an agent gets no answer, it does what any assistant does with a gap. It picks a store that answered. That is the same dynamic I described when Meta’s shopping agent was told to rank known merchants first: agents prefer merchants whose data resolves cleanly, and brand recognition does the rest.

Here is rough math, and it is hypothetical, not reported. Say your store sells a $1,800 item at 25% gross margin, so about $450 per order, and does 400 orders a month. If agent-referred sessions grow to 5% of orders, that is 20 orders. Shopify has said AI-referred sessions convert 80% better, so those are valuable shoppers. If missing freight answers cost you one in five of them, you lose 4 orders, or about $1,800 a month in gross margin. At 10% agent share and a 30% loss rate it is $5,400 a month. The thresholds are guesses, but the direction is not.

Three scenarios, with thresholds I would act on. If agent traffic is under 2% of your sessions, fix the cheap items only: policy text and labels, about a day of work. Between 2% and 8%, add structured data and real delivery dates. Above 8%, treat agent readiness like a channel, with an owner and a weekly check. Shopify’s number for your store, not mine, should set which tier you are in.

Returns are the second exposure. Only 5 of 88 stores gave a usable answer on used or open-box returns. High-ticket categories live on restocking fees, freight return costs and inspection terms. An agent that guesses wrong sets a false expectation, and the dispute lands on you. I laid out the evidence side of that in Visa’s widened chargeback evidence rules. The prevention side is in my chargeback prevention guide for high-ticket stores.

Put the same terms into your helpdesk macros, for example in Gorgias, so your human reps and the AI agents tell shoppers the same thing.

The accessibility numbers are a quiet warning. Agents read the page much like a screen reader does, so a size dropdown with no label is invisible to both. Fixing labels is cheap, free to test with WAVE, and it helps real shoppers who use assistive tech. This is also where answer-engine work pays off, which I covered in why ecommerce brands need answer engine optimization.

Dropshippers carry one more risk. Your delivery promise comes from a supplier, so your store can only state what the supplier feed gives you. If you source through Inventory Source, push the lead-time and freight fields into your product data instead of leaving them in a PDF. My dropshipping suppliers checklist covers which suppliers publish that data at all.

If you already have a store with steady orders and want someone to run this cleanup, my team handles it through the scaling service.

If wiring all of this together sounds like a second job, it is. Agent readiness now sits next to ads, supplier feeds, freight quotes and customer service on the owner’s list, and the stores that win are the ones with someone assigned to each. That is the case for letting my team build and run the store through the turnkey done-for-you service instead of doing it all yourself.

An agent asked your store about freight and heard nothing. Want my team to build and run a high-ticket store with the answers already in place? See the turnkey done-for-you service →

Six Fixes to Make Your Store Agent-Ready This Week

Do these in order. The first three take under a day.

Write your return, cancellation and open-box terms in plain sentences in Shopify admin. Shopify now writes and updates store policies automatically, so edit the generated text to state your restocking fee, freight return cost and inspection window. A Termly policy template is a fine starting draft if you want legal language to start from.
Replace the default agents.md with your real details. List shipping method by product type, freight versus parcel, lift-gate availability, return terms and a support phone number. Only five of the 89 stores serving the file had written their own, so a real file puts you in rare company.
Check /robots.txt for a Disallow on /policies/. Hunt reports 51 of 122 stores blocked it, and the block persists only on themes with an old robots.txt.liquid template, per his audit. Remove the rule and recheck, and a Semrush site audit will flag other blocked paths.
Turn on delivery dates and show a freight rate you can honor. A shipping app such as Easyship can calculate rates. A returns and tracking tool like AfterShip covers the post-purchase side that agents currently cannot touch.
Add MerchantReturnPolicy and OfferShippingDetails markup to product structured data. Google documents the fields in its merchant return policy guide. Only 7% of stores exposed merchantReturnDays, so this is where you separate from the pack. My piece on why Shopify SEO goes beyond installing apps explains why a plugin alone will not do it.
Label every size and option selector, then run WAVE on your top 10 product pages. Fix unlabeled inputs, empty links and empty buttons first, because the audit found 47%, 46% and 37% of stores had them. Book a discovery call if you want me to look at your store’s agent readiness with you.

Keep variant IDs stable and inventory current across your feeds. Wholesale feed tools such as Wholesale2b help here.

Shopify’s recent change that passes your compare-at prices to AI agents makes stale data more visible than before.

Frequently Asked Questions

Do I have to configure anything for AI agents to use my Shopify checkout?
No. Shopify says its WebMCP checkout tools do not require merchant configuration, per its September 28 changelog. Review your checkout setup anyway, because standard three-page checkout only gets the tools through Shop Pay.

Can an AI agent complete a purchase on my store without the buyer?
No. Agents must show the order and total and get the buyer’s permission before calling complete_checkout, according to Search Engine Journal’s report. Every live checkout in Hunt’s audit returned requires_escalation.

Which fix has the biggest payoff for high-ticket stores?
Delivery and freight answers. Zero of the 122 stores gave a usable shipping cost and arrival answer in the policy-tool test, and freight is the deciding question on big items. My note on LTL rates topping 80% above 2018 shows why quoting freight accurately now matters.

Does this audit prove agents are costing me sales?
No. It measures readiness, not lost orders, and it handed agents exact item IDs. Treat my dollar figures above as a hypothetical to size your own exposure.

Is the free Shopify readiness tool worth running?
Yes, as a first pass. It runs 31 checks across five categories, per Shopifreaks, but the scoring scale is not explained, so use it to find gaps and not as a pass or fail grade.

Should I add more products before fixing this?
I would not. Better answers on the products you already sell beat a wider catalog with thin data. If you are still picking categories, start with my high-ticket niches list. The free niches lead magnet goes further.

Want to work through agent readiness with other store owners and me inside the community? Join the Skool community →

Agents are going to keep asking stores questions, and the stores with written answers will keep getting the recommendation. Subscribe to the YouTube channel for daily breakdowns. More breaking news coming through the day.

If this was useful, these go deeper:

Trevor Fenner

Trevor Fenner is an ecommerce entrepreneur and the founder of Ecommerce Paradise, a platform focused on helping entrepreneurs build and scale profitable high-ticket ecommerce and dropshipping businesses. With over a decade of hands-on experience, Trevor specializes in high-ticket dropshipping strategy, niche and product selection, supplier recruiting and onboarding, Google & Bing Shopping ads, ecommerce SEO, and systems-driven automation and scaling. Through Ecommerce Paradise, he provides free education via in-depth guides like How to Start High-Ticket Dropshipping, advanced training through the High-Ticket Dropshipping Masterclass, and fully done-for-you turnkey ecommerce services for entrepreneurs who want a faster, more hands-off path to growth. Trevor is known for emphasizing sustainable, real-world ecommerce models over hype-driven tactics, helping store owners build scalable, sellable, and location-independent brands.