AI Shopping Audit 2026: How 200 UK and US Retailers Score

Everyone in eCommerce has an opinion on AI shopping agents.

Are shopping agents the future? Do people even use them? CAN people even use them?

The last question intrigued us. How many online stores will actually allow their customers to use AI agents to do their shopping for them.

The team at Blend decided to put this to the test. We’re nosy like that.

So over a week in September, we sent an AI shopping agent to 200 retailers across the UK and US to do a little shopping for us.

Our study tried to determine four key questions about the experience of an AI shopping agent:

• Can it access your site?
• Can it find the products?
• Can it read and understand them?
• Can it get a basket to the checkout?

In answering those questions, we showed what retailers are killing it when it comes to being ready for the brave new world of AI agentic shopping.

The Results at a Glance

Baby and nursery retailers came top with 88. Fashion came bottom with 49.

And 29 of the 200 wouldn't let us in at all, even when we tried TWICE, and that list includes some big hitters like Home Depot, Best Buy, Macy's, Tesco, Argos and Next.

Simple fact is, if an agent can't load your product page, it can't recommend what's on it.

It DEFINITELY can't buy it.

Subscribe to the Shopify CRO newsletter

What Is an AI Shopping Agent?

Software that does part of the shopping for someone. There are three sorts:

• The ones that answer a product question inside ChatGPT, Gemini, Copilot or Perplexity.
• The ones that open your site in a browser and click around like a customer.
• The new ones that skip the browser and talk straight to your store through a checkout endpoint. They build the basket and hand the shopper a checkout that's already filled in.

In short, it doesn’t have to go through the entire checkout process for you.

But the stores that allow one to do just that are probably more likely to make sales.

Is This Real, or Is It Hype?

Fair question, there’s a lot of AI-related guff out there at the moment.

Here's what seven independent studies from the past year say:

The one we'd look at is conversion, obviously - we are CRO experts, after all…

A year ago, according to Adobe’s numbers, a visitor from an AI tool was LESS likely to buy than your average visitor - 38% less, in fact.

This year on, they're now more likely - 42% better. That’s a hell of a swing.

The numbers actually make sense to us, from a logical point of view. These shoppers have done their comparing (and probably their umming and ahhing) before they ever turn up on your site because all these consideration conversations are happening within the AI platform itself. When they send their agent, that means a decision has probably been made.

We've written before about why AI-referred shoppers arrive warmer.

So it's still in its early stages as a movement. But it's moving fast enough that you'll probably want to know where your store stands.

What eCommerce Leaders Told US

As a precursor to this study, we also ran a survey. We asked 59 eCommerce leaders how ready they felt for AI agentic shopping and who owns that process within the business.

It seems the majority of businesses aren’t prepared for AI agent shopping, despite the fact that it’s very much already here. Businesses also don’t really have anyone leading the charge on it.

When you look into the data, our study backs that premise up perfectly.

How We Scored Each Retailer

Four questions, 17 checks:

One thing we should make clear, we never bought anything. One item in a basket, stops at the checkout page. On the endpoint test, we created one basket and one checkout, then cancelled both.

The Full Study

The full table of ten for every sector is at the end.

29 Retailers Wouldn't Let US In

That's one in seven.

By refused, we mean the store sent back an error, a challenge page or an empty page where the product page should have been. Both times we tried. Each attempt came from a residential connection in the store's own country.

Four of them refused every request we made, home page and robots.txt included. Saks Fifth Avenue, Bergdorf Goodman, Anthropologie and JD Williams. They scored 0.

We didn't expect this next bit.

Hardly any of these retailers tell AI crawlers to stay away. We could read robots.txt for 24 of the 29, and every one of those 24 allows the search crawlers from ChatGPT, Perplexity and Claude.

The refusals come from bot protection. That's the software that sits in front of the site and judges visitors by how they behave.

If you work at one of these retailers, ask your security team one question. Which AI agents are on the allow list?

Bot protection is there for good reasons. Nobody wants their website scraped without permission. Refusing can be the right call.

But who made the call? In our experience at Blend, it's usually the security or infrastructure team, doing exactly what they were asked to do, and nobody thought to ask the person who owns revenue what they'd like to happen when an AI agent turns up ready to spend a customer's money.

A few years ago, this conversation didn’t need to happen. Now the revenue team and security teams need to get in the same room and figure this problem out together.

Product Data

We could read product pages at 171 of the 200. Most have the basics. The detail is where it falls down.

A GTIN is the barcode number that identifies a product across retailers. Without it, an agent comparing prices has to guess whether two listings are the same thing. None of the beauty, fashion, luxury, pet or department store retailers we could read had one.

Shipping and returns are part of the buying decision. If they're not in the product data, the agent has to go hunting on a policy page, or leave them out.

Our co-founder Adam Pearce has written before about the vitamin and supplement brands we work with, where we found that putting detailed nutritional information on the product page builds trust, answers questions and improves conversion, and that the same structured content gets you found in ChatGPT and Gemini.

Same principle here. What helps a customer decide helps an agent decide.

It's also why product pages now have to work as landing pages for AI search. If you're after the conversion side of it, here's how we optimise Shopify product pages, and there's a free Anatomy of the Perfect Product Page to download.

Discovery was better. 107 of the 200 publish a product sitemap.

Checkout

38 of the 200 have a live agent checkout endpoint, which is a published address where an agent can search the catalogue, build a basket and ask for a checkout without needing a browser at all.

37 of the 38 are on Shopify. The other is Albee Baby.

Now, we're a Shopify agency. So yes, we would have a vested interest in pointing that out.

And to be fair to everyone else, Shopify switches this on by default. Most of those merchants didn't lift a finger.

But in our humble opinion, Shopify is the one platform that has properly optimised for the AI shopping era.

One thing most stores do have is express payment on the product page. 148 of the 171 we could read had it. Whether it helps your conversion rate is another matter.

We've seen adding it lift conversion by 35.8% for one client and removing it lift conversion by 13% for another. Test it on your own store.

Do the Big Retailers Score Worse?

On the headline numbers, yes.

We’d hazard a guess here that for the medium and big businesses, they are more likely to have the robust security teams who are stringent on bot-blocking. So big retailers are more likely to refuse.

A few years ago, that logic would have made sense. Now, we’re in a different world.

Sector League Tables

Every retailer is in the searchable table below. The tables of ten for each sector follow it.

Scores are out of 100. "Refused" means the retailer turned our automated browser away on two attempts.

What to Check on Your Own Store

TLDR version.

• Ask what your bot protection does to automated visitors. Who set the rule? When? Are named AI agents allowed through?
• Add GTIN, shipping details and returns policy to your product data. The three rarest fields in the audit. Don’t just stop there, if you have more relevant schema you could add, do it!
• Check price and availability are in the product data on every template. 41 of the 171 stores were missing price. 47 were missing availability.
• Publish a product sitemap and reference it in robots.txt. 93 of the 200 didn't have one we could find.
• Walk your own checkout as a new customer. If it forces an account, an agent stops there. So do plenty of humans.
• Check your agent checkout settings. On Shopify, look under Sales channels, then Agentic. On other platforms, ask your provider when it will support the Universal Commerce Protocol.

It's not rocket science, and none of it needs a new platform. Most of it is a conversation and a dev ticket.

At Blend, we look at conversion rate, AOV and repeat purchase together, and agent readiness goes on the same roadmap because, when you boil it down, it's one more thing that stops a customer buying.

If You're on Shopify

37 of the 200 retailers are on Shopify. They averaged 91. The lowest scored 70 and four scored 100.

So does that mean job done? No.

Shopify gives you the checkout endpoint without you doing anything, and most themes give you the basics in your product data. The rest is down to you, and most stores haven't done it.

33 of the 37 are missing at least one of GTIN, shipping or returns. 19 are missing all three.

There's a reason the same three keep coming up. Shopify's built-in structured data covers the name, brand, price and availability, but in Shopify's own documented example it doesn't include a GTIN, shipping details or a returns policy. So unless your theme or an app adds them, they aren't on the page.

Three things to do:

• Fill in the barcode on every variant. Shopify accepts a GTIN, UPC, EAN or ISBN in that field.
• Check your theme puts the barcode into the product schema. A barcode sitting in the admin is no use to an agent reading the page.
• Add shipping details and your returns policy to the product schema. Then run a product page through Google's Rich Results Test and see what comes back.

It can be done. The four stores that scored 100 (Tanners Wines, Natural Baby Shower, Babythingz and Tiger Fitness) are all on Shopify, and all four have the lot.

One more thing, because we said we'd be straight with you. The free endpoint is only part of why Shopify stores score well. Take both endpoint checks out of the scoring altogether and Shopify stores still average 89, against 66 for everyone else.

This is the sort of gap we pick up in a CRO audit at Blend, and the breakdown offer at the bottom of this report applies to you too.

How We Did It (our Methodology)

• The sample. We researched a pool of over 900 retailers across 20 sectors. Every one sells a wide catalogue, either lots of brands or its own brand at scale. We left out single-product brands, marketplaces, subscription boxes and trade-only sites. In each sector and country we split the pool into thirds by estimated monthly organic visits, then picked two large, two mid and one emerging retailer. Platform was never part of the selection.
• Where we tested from. Every request came from a residential connection in the store's own country, plus one home page request from the other country. Plain requests gave the audit's name. The browser was an unmodified Chromium, because some content delivery networks spot a name added to a browser's user agent and refuse it. We didn't impersonate a crawler or an AI company's agent at any point.
• Refusals. If a store refused the browser, we tried again from a fresh connection. Only stores that refused twice count as refusals.
• Product pages. We read up to three product pages per store in a real browser. We only counted a page once we'd confirmed it was a product page.
• Scoring. Each check carries a fixed number of points and the four layers add up to 100. If we couldn't make a check, it comes out of that store's total.

Where This Falls Short

As we said, we’re going to be honest with you about everything, that includes the limitations of this study:

• Ten retailers per sector is a small group. Sector averages describe the stores named in the tables.
• UK and US stores were tested over different kinds of connections - residential for UK and a US proxy for the US.
• It's a snapshot from 8 to 15 September 2026. Stores change settings and platforms release updates.
• Size tiers are relative to each sector's pool. An emerging department store can be bigger than a large cycling retailer.

Having said that, we believe this is an excellent snapshot of eCommerce at large and shows where improvements can be made.

Subscribe to the Shopify CRO newsletter

Frequently Asked Questions

What Is AI Shopping Agent Readiness?

How easily an AI agent can reach your store, find your products, read the product data and get a basket to the checkout. We measure those four things from outside the store, using 17 different checks.

Why Did Four Retailers Score Zero?

Saks Fifth Avenue, Bergdorf Goodman, Anthropologie and JD Williams refused every request we made, so there was nothing else to measure. We re-tested all four by hand and got the same result. The score records the refusal. It's no judgement on the store.

Are Shopify Stores More Ready for AI Agents?

Shopify stores averaged 91. Every other platform averaged 59. Part of that gap is the agent checkout endpoint, which Shopify switches on by default. Take both endpoint checks out and it's 89 against 66.

Only 12 of the 37 Shopify stores had a GTIN, though, so there's homework on Shopify too.

Which Retailers Scored 100?

Babythingz, Natural Baby Shower, Tanners Wines and Tiger Fitness. The highest scorers on other platforms were Mattressman on 92 and Real Foods on 90.

Want Your Store's Breakdown?

This report publishes scores only.

For every retailer in the audit, we hold the results of each of the 17 checks, with the evidence behind it.

If your store is one of the 200, we'll send you the full breakdown and talk you through it.

If it isn't, we can run the same audit on yours.

Not ready for that? Two other places to start:

• Benchmark your Shopify store against others in your industry. It takes about two minutes.
• See what a Blend CRO audit covers, or book a CRO mapping call.

P.S. If you're on the refused list and think we've got it wrong, tell us. We'd like to know which agents you do let in.