We Checked 24 Shopify Stores the Way AI Shopping Agents See Them

Short answer: between 6 and 14 October 2026 we checked 24 US Shopify stores using only public data: their product pages, the structured data those pages give Google, and Shopify Global Catalog, the catalogue AI shopping agents search. Every one of the 24 had at least one higher-risk mismatch, where money, orders or Google approval is at stake. In 216 everyday searches without a brand name, the store we checked was in the top 3 only 38 times and was not in the results we scanned at all 66 times.

To a shopper on the website, every one of these stores looks fine. The gaps only show when you read what Google and the catalogue read.

What did we check, and how?

• 24 stores: 17 in apparel and 7 across home and furniture, food and drink, beauty and automotive accessories. All sell in the US on Shopify. Most have between 40 and 1,000 products.
• How they were picked: from Shopify Global Catalog search results in each category, not at random. This is a small sample, not a census.
• What we read: each store's product pages, the structured data (JSON-LD) on those pages, its refund and shipping policies, and its listings in Shopify Global Catalog through Shopify's public catalogue endpoint (search_catalog at catalog.shopify.com/api/ucp/mcp).
• How we counted: each store got a report with its five biggest findings plus other observations. "N of 24" below means we recorded that issue on N stores at least once.

What did we find most often?

• 24 of 24: a core product line absent from open searches
• 22 of 24: Shopify Global Catalog's own write-up contradicting the store's pages
• 22 of 24: returns or shipping told differently to Google or shoppers than the store's policy says
• 19 of 24: live, in-stock products or sizes missing from Shopify Global Catalog
• 16 of 24: ratings that differ between the page, Google's code and Global Catalog
• 15 of 24: barcodes that aren't real GTINs
• 13 of 24: sizes or options missing in Shopify Global Catalog
• 13 of 24: a product page contradicting itself
• 13 of 24: a page title or search description copied from another product
• 9 of 24: the brand split across several names
• 6 of 24: a price on the page that checkout doesn't honour
• 3 of 24: options hidden on the page but still buyable through catalogue links

Can shoppers find these stores without typing the brand name?

Mostly not near the top. We ran 216 everyday searches that fit what each store sells (no brand name, no product name), each one 3 times in Shopify Global Catalog:

• in the top 3: 38
• 4th to 10th: 21
• 11th to 50th: 52
• past the first 50: 39
• not found at all in the results we scanned: 66

When the store was found, its median first place was 17th. 22 of the 24 stores were outside the first 50 for at least one search that fits what they sell.

Ranks move from day to day, and being listed isn't the same as being recommended. But an agent can only recommend what it finds.

Why does Shopify Global Catalog matter this much?

Because more and more agents shop through it. On 13 October 2026 Shopify launched a Shop skill for ChatGPT, Claude and other agents, with search across "over 1 billion products from millions of stores". In the US, Meta's Muse agent also searches Shopify Catalog and checks out with Shop Pay. Google's AI Mode and Gemini can now complete purchases with native checkout.

The catalogue doesn't just copy your product page. It fills gaps with its own write-up, and Shopify's own FAQ says "AI-inferred attributes may be missing or wrong." That is why we found the catalogue contradicting the store's pages at 22 of 24 stores.

What should you check on your own store first?

Start with the five that came up most and cost the most:

1. Barcodes. Every barcode should be a real GTIN with a valid check digit, unique to one product and size. A barcode that belongs to another item, or a code that isn't a GTIN at all, leaves Google and agents unsure which product is which.
2. Returns and shipping. Read your refund and shipping policy, then read what your product pages, their structured data and Merchant Center say. When they disagree, Google and agents may repeat the wrong one.
3. Unbranded searches. Search for what you sell the way a shopper would, without your brand name. Do it in an AI assistant or on shop.app, and note where you first appear.
4. The catalogue's write-up. Compare what Shopify Global Catalog says about a product (sizes, materials, ratings) with your own page. Filled-in product fields leave less for the catalogue to guess.
5. Copied titles and descriptions. Each product needs its own page title and search description. Copies make different products look identical.

What this data can and can't tell you

It shows what public product data looks like for a set of real Shopify stores in one week of October 2026. It doesn't measure sales, and it doesn't say how any one agent ranks products. Every finding was read from public pages and endpoints and dated per store.

If you want the same check run on your store, the Agentic Commerce Checklist covers what to fix first, and we're happy to run it with you.