By Steve Merrill, Founder of WRKNG Digital — August 19, 2026
How does ChatGPT decide which Shopify products to recommend?
ChatGPT recommends products it can trust and verify. That means clean feed data, valid Product structured data, retailer eligibility, and real review signals. It ranks what it can read. It skips what it can't.
I've had a hard time getting people to believe this part. Store owners think ChatGPT is crawling their pretty homepage and picking favorites. It isn't. It's reading a data layer most stores never look at.
Here's what actually happens when someone asks ChatGPT for a product.
Where does ChatGPT get product data in the first place?
ChatGPT pulls from structured product data, connected merchant feeds, and the schema markup on your pages. It does not eyeball your storefront the way a shopper does. It reads the machine-readable version of your catalog.
OpenAI rolled out Shopping inside ChatGPT and described the results as organic, chosen on relevance, not paid placement. You can read their own framing in the OpenAI Help Center. That one line matters. You can't buy your way in. You earn it with data.
So the question stops being "how do I rank" and becomes "is my data even readable."
What product feed data does ChatGPT need?
ChatGPT needs the same core fields Google Merchant Center has demanded for years: GTIN, brand, title, price, availability, condition, and product category. Miss the required ones and the product gets dropped before ranking starts.
This is where most Shopify stores lose. Not on strategy. On plumbing.
The most common gaps I see in audits:
- No GTIN. The barcode field sits empty, so the model can't match your product to anything.
- Stale availability. The feed says "in stock" while the product page says sold out.
- Price mismatch between the feed and the live page.
- Vague titles like "Blue Shirt" instead of "Men's Organic Cotton Oxford Shirt, Slim Fit, Navy."
Google publishes the field spec everyone still borrows from. If you want the exact requirements, the Google Merchant Center product data specification is the reference. Match it and you're already ahead of most stores.
How does structured data change what ChatGPT can read?
Structured data is the labeled version of your product page. Product schema tells ChatGPT the price, the availability, the rating, and the brand in a format it trusts without guessing. No schema means the model has to infer, and it usually skips what it has to infer.
Add Product structured data to every product page. Include price, availability, and aggregateRating. Shopify themes rarely ship complete markup out of the box, so check yours against Shopify's own SEO documentation and a schema validator.
We ran this on a client's store last month. Half their catalog had Product schema with no rating field and no price field. The model was reading a shell. We filled it in. Their products started showing up in test prompts within two weeks.
What is retailer eligibility, and why does it block stores?
Retailer eligibility means your store is connected through a supported feed or checkout path so ChatGPT can pull live price and stock. Without that connection, your products can be perfectly marked up and still never surface, because the model can't confirm they're buyable right now.
Think of it as the difference between being readable and being verified. Readable gets you considered. Verified gets you recommended.
OpenAI has been building toward direct checkout inside ChatGPT through its Agentic Commerce work. The stores that connect their feeds properly are the ones that get pulled into those buyable results. The stores that don't are invisible, no matter how good their product is.
How much do reviews and ratings matter?
They matter more than almost anything else once your data is clean. ChatGPT leans hard on aggregated ratings and review counts to decide what to recommend. A product with 200 reviews and a 4.6 average beats an identical product with zero reviews, every time.
This is the trust layer. The model has no hands to touch your product. So it borrows judgment from your customers.
Three things to fix here:
- Expose your real review count and average rating in your Product schema.
- Keep collecting reviews. Volume signals confidence.
- Make sure the rating on the page matches the rating in the markup.
Data does not lie. A store with thin reviews and no ratings markup tells ChatGPT one thing: skip me.
Let me bust three myths fast
Store owners keep repeating stuff that isn't true. Let me clear it.
Myth one: you can pay for placement. You can't. OpenAI says Shopping results are organic.
Myth two: a great homepage gets you recommended. ChatGPT reads feed and schema data, not your hero banner.
Myth three: this is a Google SEO trick in disguise. Some overlap, sure. But ChatGPT weighs live availability and review trust in ways classic SEO never did.
What should a Shopify store do this week?
Start with the plumbing, then build trust. Fix your feed fields first, add complete Product structured data second, confirm your retailer connection third, and pile up reviews fourth. That order matters, because ranking never gets a chance if the data is broken.
Here's the bottom line: ChatGPT recommends the products it can read, verify, and trust. Give it all three and you get pulled into answers. Miss one and you're invisible.
Most stores are sitting on broken feeds and empty schema right now. That's the opening. There's so few doing this correctly, which is exactly why it's a good time to move.
Frequently asked questions
Does ChatGPT pull products directly from my Shopify store?
Not straight from your theme. ChatGPT reads from structured product feeds, connected merchant data, and the Product structured data on your pages. Clean that data and your products become eligible to surface.
Do I have to pay to be recommended in ChatGPT Shopping?
No. OpenAI has said its Shopping results are organic and not ads. Placement comes from data quality, retailer eligibility, and review signals, not payment.
What is the single biggest reason products get left out?
Missing or broken feed fields. No GTIN, wrong price, or stale availability gets a product dropped before ranking even starts.
How much do reviews actually matter?
A lot. ChatGPT leans on aggregated ratings and review counts to decide what to trust. A product with 200 reviews and a 4.6 rating beats an identical product with none.
Ready to get your store into AI shopping results?
We build the feed, schema, and trust signals that get Shopify stores recommended inside ChatGPT. If you want your products showing up in AI answers, see how we do it at WRKNG Digital's Agentic Commerce page.
