ChatGPT Search

When a user asks a question that needs current information, ChatGPT issues web searches behind the scenes, often rewriting the prompt into several search queries through query fan-out. Retrieval draws on a Bing-derived index supplemented by OpenAI's own crawler, OAI-SearchBot, which builds the index used for search, while GPTBot collects training data separately. Retrieved pages are read, ranked, and synthesized into a single answer.

Selection favors pages that answer the question directly: clear headings, extractable passages, fresh dates, and recognizable authority. A page can rank well in classic Google yet never get cited here, because the model picks passages, not just domains.

ChatGPT Search displays inline link citations and a sources panel, and those links send measurable AI referral traffic that analytics tools see with a chatgpt.com referrer. Unlike pure chat answers from memory, search-grounded answers are auditable: you can see exactly which pages earned the citation and reverse-engineer why.

Two visibility games run in parallel: being mentioned as a recommended brand in the answer text, and being cited as a source link. Both matter, and they reward different assets, brand authority for mentions, citable content for links.

Start by allowing OAI-SearchBot in robots.txt, blocking it removes you from ChatGPT Search results entirely. Then publish content with self-contained, quotable passages and maintain strong Bing indexing, since the underlying index matters. Track which prompts in your category trigger your brand versus competitors.

Because answers vary by phrasing and day, one-off manual checks mislead. Geonimo runs your tracked prompts against ChatGPT daily and logs every mention and citation, so you can see whether your visibility is trending up or down with prompt tracking.