How ChatGPT Recommends Which Online Store to Buy From

What OpenAI has actually said, in its own words

OpenAI's help documentation on shopping with ChatGPT Search states that a product appears when ChatGPT perceives it to be relevant to the shopper's intent, and that ChatGPT considers structured metadata from first-party and third-party providers — things like price and product description — along with other third-party content such as reviews.

The documentation is also explicit about what shopping results are not: they're not paid advertisements, and OpenAI states it doesn't earn a commission on sales shown through them. That's a meaningful distinction from how many retail search results work elsewhere on the web, where paid placement is common and disclosed separately from organic results.

Trained knowledge versus live lookup

A useful distinction for understanding any AI assistant's shopping answer is where the information came from: things the model learned during training, versus information it looked up live when answering your specific question. ChatGPT's search capability — the same underlying mechanism used for shopping — retrieves current web content when a query benefits from up-to-date information, rather than relying solely on what the model learned during training.

For a store, this matters practically: a product that didn't exist when the model was trained can still be found and recommended, as long as its information is accessible for ChatGPT to retrieve live — which is exactly what structured, complete, and current product data makes possible.

What this implies for product-page clarity

Since OpenAI names structured metadata, price, reviews, and ease of use as factors it considers, the practical implication for any store — on any platform — follows directly: a product page with clear, structured, current information about price, specifications, and genuine customer reviews gives the system more to work with than a page where that information is missing, outdated, or buried in unstructured text. This mirrors Google's own long-standing guidance on product structured data, which has supported specific, itemized product attributes for years, well before AI shopping assistants existed — the underlying discipline of structuring product information clearly isn't new, even if the audience consuming it now includes an AI system.

This isn't a guess about a secret ranking formula; it's a straightforward reading of what OpenAI says it considers. A store that makes this information easy to find and keep current is working with the documented factors, not against an unknown one. Shopify's own framing of this — what it calls generative engine optimization — names nearly the same list: data completeness, structured formatting, and trust signals, independently arriving at a similar conclusion from the merchant side of the relationship rather than the platform side.

What remains genuinely undocumented

It's just as important to be clear about what OpenAI has not published. There's no public, exact formula for how heavily each factor is weighted relative to the others, no disclosed threshold for how many reviews is "enough," and no official statement about how quickly changes to a product page affect what ChatGPT shows. Anyone — including a well-meaning guide like this one — who states those specifics as confirmed fact is presenting a guess as documentation.

The responsible approach is to act on what's actually been said — relevance, structured metadata, price, reviews, ease of use — and treat everything more specific than that as a reasonable inference, not a guarantee.

FAQ