-
Written by:
Asad Abdullah Zafar
- Published
- 9 min read
- Shopify
Quick answer: On 25 October 2026 we checked 162 well-known Shopify brands against what an AI shopping agent reads. None of them used robots.txt to block AI search or shopping crawlers from their product pages, and 94% served an llms.txt file, mostly the one Shopify now writes for every store. But the product data agents need to compare and recommend was thin: 40% of stores gave a GTIN or barcode, 12% gave shipping details and 14% a return policy. Only 39% gave price, stock status and a GTIN together.
ChatGPT, Perplexity, Google and Shopify's own agents now find products, compare them and, increasingly, check out on a shopper's behalf. That changes what a product page is for. It still has to persuade a person, but first it has to be read by a machine that decides whether the product is even on the shortlist.
So we asked a simple question: if an AI agent visits a well-known Shopify store today, is it allowed in, and does it find what it needs? This report sets out what we found, how we checked and what store owners can fix.
A disclosure first. Loop2Tech is a Shopify studio, and we sell the kind of work this report points to. We report only what each store's public pages say, and you can check every row in the full dataset (CSV).
How we checked
We started from a list of 271 direct-to-consumer brands widely reported to sell on Shopify, across fashion, beauty, home, food, wellness and electronics. On 25 October 2026 a script visited each store, identifying itself as Loop2TechResearch and pausing between stores. For each store it:
- confirmed the store runs on Shopify by reading its public product list, and recorded why when it couldn't;
- read robots.txt and tested 12 AI crawlers against a real product page: six that power AI search and shopping answers (OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User, Claude-SearchBot and Claude-User) and six used for model training (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, CCBot and Amazonbot);
- opened up to three ordinary, in-stock products, skipping samples, gift cards, bundles and sets, and read the product structured data (JSON-LD) on each page;
- checked whether the store serves an llms.txt file, and whether it is Shopify's generated file or one the merchant wrote.
A store counts as having a piece of product data when most of the product pages we read carry it, so one odd page doesn't decide the result.
162 stores could be audited and we read product pages on 156 of them. We excluded 109: 71 no longer answer like a Shopify storefront, 29 are on Shopify but block the public product list, 6 couldn't be reached and 3 had no ordinary product to sample. The excluded stores and the reason for each are in the dataset.
Three limits matter. robots.txt is only one way to block a crawler; a firewall can turn bots away without saying so, and we can't see that from outside. We read JSON-LD structured data, which is what Shopify themes use, and not older microdata. And these are well-known brands with in-house teams, so smaller stores are likely to score lower, not higher.
Finding 1: no store keeps AI shopping agents out
| What robots.txt does to product pages | Stores | Share |
|---|---|---|
| Blocks at least one AI search or shopping crawler | 0 | 0% |
| Blocks at least one AI training crawler | 2 | 1% |
| Serves a robots.txt at all | 160 | 99% |
Not one of the 162 stores blocked ChatGPT's, Perplexity's or Claude's search and shopping crawlers from a product page. Only two blocked any training crawler: Bulletproof blocked GPTBot, ClaudeBot, Applebot-Extended and CCBot, and Coway's Airmega store blocked Amazonbot.
That is largely Shopify's doing. Its default robots.txt keeps crawlers out of carts, checkouts and account pages but leaves product pages open, and almost nobody changes it. For brands that want to be recommended by AI assistants, that default is the right one. Blocking is not the problem.
Finding 2: Shopify is already inviting agents in
153 of the 162 stores (94%) served an llms.txt file, a plain-text file that tells AI agents what a site is and how to use it. For 135 stores (83%), it was the "Agent Instructions" file Shopify now generates automatically, which points agents to Shopify's Shop skill and checkout.
Only 18 stores (11%) had written their own, including Away, Tatcha, Liquid I.V., Necessaire, Thursday Boot and Tecovas. Nine served none.
In practice, most of these brands have an agent-facing front door they didn't build and may not know about. That is useful, but the generated file describes the platform's tools, not the brand's products, policies or what makes them different.
Finding 3: the basics are there, the deciding details aren't
Across the 156 stores whose product pages we read:
| Product data in the page's structured data | Stores | Share |
|---|---|---|
| Product structured data of any kind | 147 | 94% |
| Price | 145 | 93% |
| Brand | 145 | 93% |
| Stock availability | 142 | 91% |
| SKU | 119 | 76% |
| GTIN or barcode | 63 | 40% |
| Review rating | 38 | 24% |
| Return policy | 22 | 14% |
| Shipping details | 18 | 12% |
Price, brand and stock status come from Shopify's themes, so nearly everyone has them. The fields that drop away are the ones a merchant has to add: barcodes, reviews in the structured data, and shipping and return terms.
Those are the fields an agent uses to make a decision. A GTIN is how it knows that the shoe on your store is the same shoe on three others. Ratings are how it judges quality without reading every review. Shipping cost, delivery time and returns are often the first questions a shopper asks an assistant, and the answer has to come from somewhere.
Only 39% of stores (61) gave price, availability and a GTIN together, the minimum for an agent to match a product across sellers and know it can be bought. Only 16 stores (10%) gave both shipping and return details, and just 3 gave price, availability, GTIN, rating, shipping and returns all at once.
Finding 4: nine big brands give agents no product data at all
Nine stores had no product structured data on the pages we read: Gymshark, Brooklinen, Jeffree Star Cosmetics, Merit Beauty, Function of Beauty, Snowe, Grove Collaborative, Set Active and Dagne Dover. Some run custom or headless storefronts, where structured data has to be added by hand and is easy to lose in a rebuild.
An agent can still read the visible page, but it has to guess at price, stock and variants instead of being told. When an assistant is choosing between two similar products, the one it doesn't have to guess about is the safer recommendation.
Product descriptions in the structured data were short, too: a median of 37 words. That is enough to name a product, but rarely enough to answer "is this good for wide feet?" or "will this work on a gas hob?"
What store owners should fix first
- Add barcodes to your variants. In Shopify admin, each variant has a Barcode field. Many themes output it as a GTIN once it is filled in; check a product page to confirm yours does. If you sell your own brand, register GTINs with GS1.
- Put shipping and return terms in your structured data. Google reads
shippingDetailsandhasMerchantReturnPolicyfor its shopping results, and any agent reading the page can use the same fields. Few themes add them by default, so this usually needs an app or a theme edit. - Get review ratings into the product schema. If your reviews app shows stars on the page but not in the JSON-LD, agents may not see them.
- Write your own llms.txt. Shopify's file is a good start. Add what only you can say: your best sellers, sizing advice, shipping regions and return terms.
- Check after every theme change or rebuild. Headless and heavily customized stores are the most likely to lose structured data without anyone noticing.
Our free website audit tool checks a store's structured data and AI crawler access in a couple of minutes.
Use the data
The full dataset lists every store we checked, with the product URLs we read, which crawlers robots.txt blocks, each structured data check, the llms.txt result, the reason for any exclusion and the time of the check. You are welcome to use it; please link to this page as the source.
If you would like this fixed on your own store, our Shopify development service covers structured data and theme work, and our AI visibility audit shows how AI assistants currently describe your brand and products. Send us your store URL and we will tell you what an agent sees today.
Frequently asked questions
Do Shopify stores block AI shopping agents?
Almost never. In October 2026, none of 162 well-known Shopify brands used robots.txt to block AI search or shopping crawlers such as OAI-SearchBot, PerplexityBot or Claude-SearchBot from product pages, and only 2 blocked any AI training crawler.
Does Shopify create an llms.txt file automatically?
Yes. Shopify now serves an "Agent Instructions" llms.txt for stores that haven't written their own, pointing AI agents to its Shop skill and checkout. In our October 2026 check, 83% of 162 Shopify brands served Shopify's generated file and 11% had written their own.
What product data do AI shopping agents need?
At minimum, a price, stock availability and a GTIN or barcode, so the agent can match the product across sellers and know it can be bought. Ratings, shipping details and return policy help it recommend the product. Only 39% of the Shopify brands we checked gave price, availability and a GTIN together.
How do I add a GTIN to my Shopify products?
Fill in the Barcode field on each product variant in Shopify admin. Many Shopify themes then output it in the product's structured data as a GTIN. Check a product page with Google's Rich Results Test to confirm it appears.
Is my Shopify store ready for AI shopping?
If your product pages carry price, availability, GTIN, ratings, shipping and return details in their structured data, and you serve an llms.txt that describes your products, you are ahead of most well-known brands. Just 3 of the 156 stores we read had all six product fields.
