How to rank in ChatGPT is one of the most searched questions in SEO right now, and most of the answers to it are invented. Lists of “ChatGPT ranking factors” circulate with no source behind them. The documented part is much smaller and much more useful: OpenAI publishes which crawlers it runs, what each one is for, and what happens if you block them. Start there, because getting it wrong can remove you from ChatGPT search entirely.
The three OpenAI crawlers, and which one matters
OpenAI’s overview of its crawlers lists separate user agents with separate jobs. Confusing them is the most common mistake site owners make.
| User agent | What OpenAI says it does | If you block it |
|---|---|---|
| OAI-SearchBot | Used “to surface websites in search results in ChatGPT’s search features”. Not used for training | You “will not be shown in ChatGPT search answers”, though you can still appear as navigational links |
| GPTBot | Crawls content “that may be used in training our generative AI foundation models” | Your content is opted out of training; search visibility is governed by OAI-SearchBot instead |
| ChatGPT-User | Used for certain user actions, such as visiting a page when a user asks ChatGPT or a Custom GPT a question | Those user-triggered visits are affected, not search crawling |
The key line is the third column of the first row. If OAI-SearchBot is blocked, the content of your pages will not be shown in ChatGPT search answers. No amount of content work fixes that.
Note the exception, too. A blocked site can still appear as a navigational link, for example when a user asks for a brand by name. That is a long way from being cited as a source. If ChatGPT only ever links to your homepage when people ask for you by name, check whether OAI-SearchBot is being turned away.
ChatGPT-User is a different case again. It visits pages because a user asked for something in a conversation, not to build a search index. Blocking it affects those live visits, and it is worth knowing which of the three agents shows up in your logs before you write any rules.
How to rank in ChatGPT: allow search, decide on training
Because the two crawlers are separate, you can appear in ChatGPT search while opting out of model training. A robots.txt that does both looks like this:
Whether to block GPTBot is a business decision, not an SEO one. Blocking it keeps your content out of future training data. Allowing it does not, on anything OpenAI has published, improve your search visibility. Make the call on its merits.
OpenAI says it can take around 24 hours from a robots.txt update for its systems to adjust, so do not panic if a change does not show up in the first hour. Our guide to robots.txt for SEO covers syntax, testing and the mistakes that block more than you meant to.
Check what is actually blocking you
Many sites block OAI-SearchBot without meaning to. The usual causes:
- A blanket AI block. Someone added every AI user agent to robots.txt in one go, search crawler included.
- A CDN or firewall rule. Bot protection that challenges or blocks unfamiliar crawlers can stop OAI-SearchBot before robots.txt is even read.
- A plugin setting. Some security and SEO plugins ship AI-blocking toggles that do not distinguish between training and search crawlers.
Check robots.txt by hand, then check your server or CDN logs for requests from OAI-SearchBot and the responses they received. A crawler that only ever gets a 403 is blocked, whatever robots.txt says.
Allowing the crawler is a requirement, not a boost. Nothing OpenAI publishes suggests that being crawlable earns a better position than any other crawlable site. It only puts you in the pool.
What happens after access: no published ranking factors
Once OAI-SearchBot can reach you, OpenAI’s documentation stops. It does not publish how ChatGPT search chooses which pages to cite, and we will not invent a list. Treat any article that ranks “the top ChatGPT ranking factors” with suspicion unless every item links to OpenAI.
What you can reasonably do is the work that makes a page worth citing in any answer engine:
- Answer the question in the first lines of the relevant section, plainly.
- Back claims with specifics and sources a reader could check.
- Cover the follow-up questions a user would ask next.
- Keep pages fast, crawlable and free of content hidden behind scripts or logins.
That is ordinary SEO, which is the point. Our comparison of GEO vs SEO sets out why most AI-answer advice collapses into the same work, and why the research behind “generative engine optimisation” does not support the bigger claims made for it.
Tracking ChatGPT traffic
OpenAI’s publisher FAQ says ChatGPT adds utm_source=chatgpt.com to referral links. That makes the traffic visible in analytics: filter sessions by that source and you can see which pages ChatGPT sends people to, how many arrive and what they do once they land.
This is the most honest measure available. It will not tell you how often you were cited without a click, but it tells you whether ChatGPT search is worth your time. Our post on Search Console vs Google Analytics covers where referral data lives and what each tool can and cannot report.
The limitation: you cannot see the answers
ChatGPT answers vary by user, conversation and wording. There is no equivalent of a rankings report, and nothing in OpenAI’s crawler documentation offers publishers impression data for search answers in the way Search Console does for Google. Third-party tools that sample prompts can give a rough picture, but their prompts are not your audience’s prompts. Referral traffic is the hard number; everything else is an estimate.
So the plan is short. Allow OAI-SearchBot and confirm in your logs that it is getting through. Decide on GPTBot as a separate, business question. Tag and watch chatgpt.com referrals for a few months before you decide how much effort the channel deserves. Put the rest of your time into pages that answer real questions well, because that is the only lever left once access is sorted.
Skip the extras that promise more than they deliver. A file such as llms.txt has no documented effect on ChatGPT search, as our post on what llms.txt is explains, and “AI-optimised” rewrites of pages that already rank are a solution looking for a problem. The algorithm archive collects our other posts on AI search.
Frequently asked questions
Which OpenAI crawler do I need to allow for ChatGPT search?
OAI-SearchBot. OpenAI says sites opted out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links.
Does blocking GPTBot remove me from ChatGPT search?
No. GPTBot crawls content that may be used for training. Search visibility is governed by OAI-SearchBot, so you can allow one and block the other.
How long does a robots.txt change take to apply?
OpenAI says it can take around 24 hours from a site’s robots.txt update for its systems to adjust.
How do I track traffic from ChatGPT?
OpenAI says ChatGPT adds utm_source=chatgpt.com to referral links, so you can filter for that source in your analytics.
The takeaway Allow OAI-SearchBot, decide on GPTBot separately, and track chatgpt.com referrals. OpenAI publishes no ranking factors, so the rest is ordinary, citable content.
