OpenAI documents two sources. The first is its own index: the OpenAI crawler page says "OAI-SearchBot is for search" and that a site which blocks it "will not be shown in ChatGPT search answers". The second is Bing. OpenAI's search documentation names Bing and Shopify as its third-party search providers, and its enterprise help page says Bing is the only third-party provider for Enterprise and Edu workspaces, both quoted in a Semrush article from August 2025.
Google reaches ChatGPT too, at least sometimes. In that Semrush piece, Abhishek Iyer made a page about an invented term that only Google could crawl, and ChatGPT cited it. Leigh McKenzie repeated the test and found the paid ChatGPT version did this and the free one did not. Each test used one page, and the article does not say how often this happens, so treat it as a demonstration rather than a rate.
What we measured
On 2026-10-21 we asked ChatGPT 25 buyer questions about AI visibility through its API with search switched on. It cited 47 sources in total, 43 of them pages owned by OpenAI, Google or Perplexity. On 7 of the 25 questions it cited nothing at all. Every link it returned carried a utm_source=openai parameter. The full list is on the What ChatGPT cites page.
So the index question matters less than the selection. ChatGPT searches, but it cites few pages and prefers the platform's own documentation whenever one exists.
What to do
Allow OAI-SearchBot in robots.txt. Make sure Bing has indexed your site, through Bing Webmaster Tools and IndexNow, since Bing is the one provider OpenAI names. Keep Google indexing healthy for the cases where ChatGPT falls back to it. Then write the page ChatGPT would cite if no official document answered the question: a direct answer with dated facts.
