Why ChatGPT can't see your Shopify products
Not a bug, not a ban, and not usually a Shopify setting. Most of the time it's an incomplete product record. Here's the actual mechanism, and the fields that matter.
Check any Shopify store — free, no install
Reads the store’s public product feed and runs 15 of our 19 checks — no login, no install, nothing in the store is changed. Four checks need access to the store and are marked as not run rather than counted against you.
The short version. Shopify syndicates a structured feed of your product data — title, description, options, images, price and availability — to AI shopping assistants, under what it calls Shopify Catalog and Agentic Storefronts. A shopper asking ChatGPT or Copilot a question doesn't browse your store; the assistant matches the question against structured fields in that feed. A product with a blank description, no category, no real option names or no usable price has nothing for the assistant to match against, even if the product itself is published and for sale.
The mechanism, plainly
An AI shopping assistant doesn't read your storefront the way a person does. When a shopper asks "find me a waterproof jacket under $150," the assistant isn't scanning product photos or reading marketing copy — it's running that question against structured data: category, price, attributes like material and fit, and the description as text it can search. Shopify's own documentation describes Shopify Catalog as syndicating "title, description, options, images, price and availability" specifically "in a way that AI agents can parse and understand" — the phrase "parse and understand" is the important part. A product record can be true and complete for a human shopper and still be unparseable for a matching system if the facts a question would need aren't in a field the system reads.
This is also why a product can be published, in stock, and perfectly visible on your storefront while still never surfacing in an AI assistant's answer. Being "in your store" and being legible to Shopify Catalog are two different states, and the gap between them is almost always product data, not a setting you flipped by accident.
The fields that actually break the match
No description, or a one-line one · critical
The description is the main body of text an assistant can search. An empty description means there's almost nothing to match a shopper's question against beyond the title. A one-line description is barely better, because it rarely contains the facts a shopping question actually asks about — material, dimensions, what it fits, what's in the box.
A useful test: pick a product and ask "is this waterproof?", "will it fit a 15-inch laptop?", "is it machine washable?" If the answer isn't written in the description field, it doesn't exist as far as a matching system is concerned, no matter how obvious it is from the photo.
No product category · critical
Shopify's product taxonomy is the vocabulary that tells a matching system a "runner" is a rug and not a shoe. A product with no category assigned is, structurally, an unlabelled item — and category is also what determines which structured attributes (material, fit, capacity, and so on) Shopify offers for that product at all. Skip the category and you skip access to most of the other structured fields too.
No usable price · critical
Price is a filter in nearly every real shopping question — "under $150," "around $50." A product with no price, a zero price, or a price only some hidden variant carries is excluded from any question that filters on price, which is most of them.
Vague or missing attributes · major
Once a category is set, Shopify surfaces category-specific attribute fields — colour, material, fit, capacity, and so on, depending on the category. These are what a specific shopping question is actually matched against: "in blue," "size 10," "leather." A product with a category but no attributes filled in is searchable by name only, which fails any question with a specific requirement in it.
Option names left as "Option 1" · major
If variants were added without renaming the option, Shopify stores the values correctly
(Blue, Red, Large) but labels the option itself Option 1, Option 2.
The values are right but nothing says what kind of thing they are, so a question like "do
you have it in blue, size 10?" can't be matched even though the data technically contains
both facts.
No images, or images with no alt text · minor
Alt text is the only description of an image a system that isn't doing image recognition can read. It's a smaller factor than the ones above, but it's one of the cheapest to fix, and Shopify names image coverage as one of the signals behind the listing-quality indicator it shows for products already in Catalog.
What this is not
This isn't a claim that fixing these fields guarantees your product gets recommended by any specific assistant. Ranking and selection inside ChatGPT, Copilot or any other AI shopping surface is that platform's own logic, and none of it is published in enough detail for anyone outside those companies to promise a placement. What's checkable, and what this page is about, is the floor: a product missing these fields is structurally unable to match most shopping questions, independent of ranking. Fixing the fields removes that ceiling. It doesn't buy a result above it.
It's also not usually a Shopify setting working against you. If you've already confirmed Agentic Storefronts or Shopify Catalog access is switched on for the AI platform in question, see our page on Agentic Storefronts and how to control it for the toggles that actually exist. Most of the time the setting is fine and the product record is the gap.
If you have more than a handful of products
Checking each of these fields one product at a time works for a small catalog. It stops being reasonable somewhere past a few dozen products, because nothing in the Shopify admin will show you, across your whole catalog, which products have a thin description versus an empty one, or which still say "Option 1." That's what we built AI Catalog Readiness for — it scans a catalog against checks like these and ranks what to fix first by the revenue behind each product. It's read-only and it diagnoses; it doesn't control or guarantee where any AI assistant places a product, and our terms of service say so plainly.
Run the scan on your own store
One field, 30 seconds, nothing installed
Reads the store’s public product feed and runs 15 of our 19 checks — no login, no install, nothing in the store is changed. Four checks need access to the store and are marked as not run rather than counted against you.
If you would rather have this run across your whole catalog and ranked by the revenue behind each product, that is AI Catalog Readiness — read-only, free for your first 50 products, on the Shopify App Store.