ChatGPT SEO: how to appear in ChatGPT search

OpenAI’s help page on web search puts it in two steps: allow OAI-SearchBot to crawl the site, and confirm that the host or CDN allows traffic from OpenAI’s published search bot addresses (Searching the web with ChatGPT). The same page says ChatGPT ranks results using multiple factors meant to help people find relevant, reliable information, and that placement is not guaranteed.

Its crawler page adds the consequences (Overview of OpenAI Crawlers): a site that keeps OAI-SearchBot out is not shown in ChatGPT search answers, though it can still appear as a navigational link, and after a change to robots.txt it can take about 24 hours for OpenAI’s systems to follow.

• OAI-SearchBot finds pages for ChatGPT’s search. This is the one ChatGPT SEO is about: OAI-SearchBot and ChatGPT-User.
• ChatGPT-User opens a page when someone asks ChatGPT, or a custom GPT, something that needs it. OpenAI says it doesn’t crawl the web on its own, and that because a person starts these visits, robots.txt rules may not apply.
• GPTBot collects pages that may be used to train OpenAI’s models. Blocking it says your content should not be used for training: GPTBot.

OpenAI says each setting is independent of the others. So you can keep training out and stay in search, and a robots.txt that does exactly that reads:

User-agent: GPTBot
Disallow: /

User-agent: OAI-SearchBot
Allow: /

We measured it. In our study of robots.txt files, of the 1,331 sites that block GPTBot by name, 17.1% also block OAI-SearchBot; among the ones that wrote the rule themselves rather than using Cloudflare’s managed file, 31.3% (Block training, keep answers?). The usual ways it happens:

• One group for every AI bot. A list of AI user agents over a single Disallow: /, copied to stop training, with OAI-SearchBot in it too.
• A wildcard that closes everything. User-agent: * with Disallow: /, and groups that let in only Googlebot and Bingbot. OAI-SearchBot has no group of its own, so it reads the * one and stays out.
• The edge, not the file. A firewall rule, a bot-protection setting or Cloudflare’s AI bot settings can refuse it before robots.txt has a say. OpenAI publishes the addresses to let through, searchbot.json; on Cloudflare: Cloudflare and SEO.

Paste your robots.txt into the robots.txt validator to see the line that decides for OAI-SearchBot. The Trakis audit reports it as robots.txt keeps AI search out, with how many pages each bot is kept out of.

OpenAI says ChatGPT search sometimes partners with other search providers: it rewrites the question into one or more targeted queries and sends them to those providers, then may send more specific ones after reading the first results. Among the providers’ privacy policies it points to is Microsoft’s (Searching the web with ChatGPT). It doesn’t say which provider answers which question.

Our advice, not OpenAI’s: Microsoft runs Bing, so make sure Bing knows your pages too. Bing Webmaster Tools shows what Bing has indexed, and IndexNow tells it when a page changes. A site that works for search engines in general works here too: pages that answer, at addresses that work, in a sitemap.

OpenAI publishes no list of ranking factors, and the two pages above say nothing about JavaScript or structured data. What a page can do is be readable by any crawler and say what it is about early. These are the audit’s AI readiness checks; each page says why, and where the evidence stops:

• The content in the HTML the server sends, not drawn later by scripts: whether OpenAI’s crawler runs JavaScript isn’t said on the pages above, and HTML is read either way. Content only appears through JavaScript.
• A short answer right under the heading, so the page says what it offers before the reader scrolls: No short answer under the heading.
• Articles with a named author: Article without a named author.

On llms.txt: neither page says OpenAI’s crawlers read one. The crawler page links an llms.txt only as the index of OpenAI’s own documentation. What the file is, and who says they read it: llms.txt, explained.

• robots.txt: open https://your-domain/robots.txt, the file crawlers actually get, and run it through the validator for OAI-SearchBot.
• The edge: look for firewall, bot or AI-crawler rules at your host or CDN that could refuse OpenAI’s addresses.
• Your logs: requests with OAI-SearchBot in the user agent, from an address in searchbot.json, mean it is reading the site. A user agent alone can be faked; the address is what proves it.
• ChatGPT itself: ask, with search on, a question one of your pages answers well, and see whether it cites you. Placement isn’t guaranteed, so one answer proves little either way.

Does blocking GPTBot keep my site out of ChatGPT search? Not by itself: OpenAI says the settings are independent, and search is OAI-SearchBot. But a rule written to block GPTBot often catches OAI-SearchBot too; check the file.

How long until ChatGPT sees a robots.txt change? OpenAI says about 24 hours for its systems to adjust. Being found again after that is up to its ranking.

Is ChatGPT SEO different from SEO? Mostly not. The differences are a crawler of its own to let in, search partners to be found by, and answers that cite their sources inside the text instead of a page of links.

Can I see whether ChatGPT sends me visitors? In your analytics, look for visits referred from chatgpt.com.

OpenAI’s pages, read on 4 November 2026.

• OpenAI, Searching the web with ChatGPT: what makes a site eligible, ranking and placement, search partners and rewritten queries.
• OpenAI, Overview of OpenAI Crawlers: OAI-SearchBot, ChatGPT-User and GPTBot, opting out, the 24 hours, independent settings.
• OpenAI, OAI-SearchBot IP addresses (JSON).
• Trakis, Block training, keep answers?: the share of sites that block OAI-SearchBot along with GPTBot.