Surfaces · Search-grounded answers

How ChatGPT reads your catalog

ChatGPT answers by grounding the question in live web search, then reading the pages it finds. It sees your site the way a crawler does — the rendered page and its structured data — not the way a customer does.

What it ranks on

• Crawler access — GPTBot and OAI-SearchBot must be allowed in robots.txt.
• Complete JSON-LD for whatever you offer: Product with offers and price, Service with an area served, LocalBusiness with address and hours, SoftwareApplication with plans.
• Something confirmable — a GTIN or MPN on a product, a licence or registration on a firm, a verifiable address on a local business.
• Titles and descriptions that answer the customer's actual question.

What breaks

• robots.txt turns the crawlers away — nothing can be read.
• What you offer lives only in prose, or only in JavaScript the crawler doesn't execute.
• Nothing the model can confirm you by, so it names someone it can verify.
• Thin headings that omit the thing the customer actually typed.