Skip to content
SiteFaults

How to rank on ChatGPT search

The three OpenAI crawlers and which to allow, why being in Bing's index matters, and what makes ChatGPT cite your page over a competitor's.

By Max7 min read

“Ranking on ChatGPT” is not really ranking — there is no numbered results page. What you are actually after is being cited: named and linked when ChatGPT answers a question in your field. That happens through a different pipe from Google’s AI Overviews, and getting the plumbing right is most of the battle. Here is how ChatGPT finds sources, and what makes it pick yours.

How ChatGPT gets its answers

ChatGPT draws on three different mechanisms, and confusing them is why so much advice on this is wrong:

  • Its trained knowledge — what the model absorbed during training, with a cutoff date. You cannot directly influence this for a specific query, and it is not where citations come from.
  • ChatGPT search — the live search feature that fetches current results and shows sources with links. This is what you are optimising for, and it relies on an index built by OpenAI’s own crawler, which has also drawn on Bing’s search results.
  • On-demand browsing — when a prompt makes ChatGPT fetch a specific URL then and there.

Each of those last two uses a different crawler, and telling them apart is the single most useful thing to understand here.

Step one: allow the right OpenAI crawlers

OpenAI operates three separate user agents, and they do different jobs. If you want to be cited in ChatGPT search, the one that matters is OAI-SearchBot:

  • OAI-SearchBot — builds the index behind ChatGPT search. Allow this one to be eligible for citations.
  • ChatGPT-User — fetches a page live when a user’s prompt triggers a browse. Allow it so those fetches succeed.
  • GPTBot — used for training, not search. Blocking GPTBot does not remove you from ChatGPT search results; allowing it does not put you in them. This is the crawler most “block ChatGPT” guides mean, and it is the wrong one for this goal.

So the robots.txt that keeps you eligible for ChatGPT search looks like this — note that you can allow the search crawlers while still opting out of training if you want to:

# Eligible for ChatGPT search citations:
User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

# Optional: opt out of training only, without
# affecting search citations:
User-agent: GPTBot
Disallow: /

Check what you currently allow with the robots.txt checker — a surprising number of sites block all three by accident, inherited from a plugin or CDN default nobody chose.

Step two: be in Bing’s index

Because ChatGPT search has drawn on Bing’s results, being indexed by Bing is a practical advantage that most people optimising “for AI” overlook entirely. The steps mirror Google’s: verify your site in Bing Webmaster Tools, submit a sitemap, and confirm your key pages are actually indexed. If you have never looked at Bing, this is often the fastest available win — and the general mechanics are the same as getting indexed on Google, just on a second search engine.

Step three: give it something worth quoting

Access gets you eligible; content gets you chosen. ChatGPT synthesises an answer and attaches the sources it leaned on, and it favours pages that make a clear, checkable claim it can stand behind — the same writing discipline that underpins generative engine optimization across every AI engine:

  • Lead with the answer. A self-contained passage that states the fact directly, near a descriptive heading, is far more quotable than the same fact buried three paragraphs in.
  • Be specific and current. Concrete numbers, dates, steps and definitions are what a summary is built from. Keep them up to date — models weight freshness for anything time-sensitive.
  • Earn corroboration. ChatGPT leans toward claims that are consistent across multiple reputable sources. Being mentioned and linked elsewhere — the ordinary work of digital PR and being genuinely useful — makes your version of a fact the one it trusts.
  • Be clear about who you are. A consistent description of your organisation and expertise, backed by structured data, helps the model attribute a statement to a credible named source rather than an anonymous page.

What does not move the needle

There is no ChatGPT submission form and no markup that requests a citation. An llms.txt file will not do it either — it is a proposal no major AI provider has confirmed it reads, and it is honestly not where your effort belongs. Keyword density, hidden text and the other tricks from a decade ago are, if anything, counterproductive: they make a passage less quotable, not more.

How to check where you stand

The direct test is to ask ChatGPT the questions your customers ask and see whether it cites you, a competitor, or nobody in your niche. Do it periodically — results shift as the index updates. For the underlying readiness — crawler access, structured data, and whether your pages carry quotable answers at all — run the website GEO check, and see what an AI visibility checker tests for how that assessment works. The companion guide, how to appear in Google AI Overviews, covers the engine that works the opposite way — on Google’s own index rather than OpenAI’s.

Frequently asked questions

How do I get my website cited by ChatGPT?
First make sure ChatGPT's search crawler can reach you: allow OAI-SearchBot and ChatGPT-User in robots.txt. Then get indexed by Bing, since ChatGPT search has drawn on Bing's results. Finally, give it quotable content — clear, self-contained answers near descriptive headings, with specific facts it can stand behind. There is no submission form and no citation guarantee.
Does blocking GPTBot stop ChatGPT from citing my site?
No. GPTBot is OpenAI's training crawler, not its search crawler. Blocking GPTBot only affects whether your content is used to train models — it does not remove you from ChatGPT search results. The crawler that matters for citations is OAI-SearchBot, with ChatGPT-User for live fetches. You can block GPTBot to opt out of training while still allowing the search crawlers.
Does ChatGPT use Google's search index?
No. ChatGPT search relies on an index built by OpenAI's own crawler and has drawn on Bing's search results, not Google's. That is why the tactics differ from Google AI Overviews: for ChatGPT, being indexed by Bing and allowing OpenAI's search crawler matter, whereas Overviews run entirely on Google's index.
Will an llms.txt file help me rank on ChatGPT?
There is no evidence it will. llms.txt is a proposed file for listing your key pages to AI tools, but no major provider — OpenAI included — has confirmed it reads or acts on it. Your effort is far better spent on crawler access, being indexed by Bing, and writing genuinely quotable content. Treat llms.txt as optional and low-priority.
Why does ChatGPT cite my competitor instead of me?
Usually one of three reasons: the competitor is indexed by the search engines ChatGPT relies on and you are not; their page states the answer more directly and quotably than yours; or their claim is corroborated across more reputable sources. Check your crawler access and Bing indexing first, then tighten the specific passage that should answer the question.
How often should I check my ChatGPT visibility?
Check periodically — roughly monthly — because the underlying index updates and answers shift over time. Ask ChatGPT the real questions your customers ask and note whether it cites you, a competitor, or nobody. Pair that with a technical readiness check of crawler access and structured data so you know whether a gap is access, content, or simply that the index hasn't caught up yet.

Run a free check on your own site

Paste any URL and get your score, your Core Web Vitals and every issue we find in about 30 seconds. No signup, no card.

Free, no signup, no card. Results in about 30 seconds, and your report is deleted automatically after 30 days.