How Do You Get Your Brand Cited by ChatGPT?

You can’t guarantee a ChatGPT citation, but you can make your site eligible and worth citing. OpenAI says any public website can appear in ChatGPT search, so long as you aren’t blocking its search crawler, and it doesn’t publish what decides who gets linked.

That gap between “eligible” and “chosen” is where most of the bad advice lives. This guide sticks to what OpenAI has documented, shows you how to check your own setup, marks what’s still unknown and finishes with steps that cost little and carry no downside.

We read OpenAI’s pages on 3 October 2026, and its help articles were marked as updated within the previous month. If you want a baseline of what ChatGPT says about you before you change anything, our AI visibility audit is built for that.

How Does ChatGPT Search Work?

ChatGPT may search the web when a question would benefit from current information, typically rewrites the question into targeted queries, and answers with citations that link to sources. That’s the process OpenAI describes in its help article on searching the web.

Stage	What OpenAI says	What it means for you
Decide to search	ChatGPT may search automatically when a question would benefit from current information	Fact-seeking and current-information prompts are where your page can appear
Rewrite the query	It typically rewrites the question into one or more targeted queries, and may add more specific ones after reviewing results	The search that finds you may differ from the prompt the user typed
Retrieve	It sometimes partners with other search providers	Being indexed by those providers matters
Answer	Responses that use web search may include citations	A response with no search has no live citation to win

When does ChatGPT search?

Either automatically, or because the user starts a search manually from the tools menu. OpenAI describes citations only for responses that use web search, so a reply written without searching has no live link to your page.

How does it build the search query?

OpenAI’s own example turns a question about drugs targeting CCR8 for cancer into a query for “CCR8 immunotherapy drug development 2025”. After reviewing the results, it may send a more specific follow-up query.

That’s query fan-out in practice, and it means you compete on narrower phrasing than the head term. If memory is enabled, ChatGPT may also use relevant saved memories when rewriting a query.

Which search providers does it use?

The help article links the privacy statements of Microsoft and Shopify when it explains information shared with search providers. OpenAI’s launch post of 31 October 2024 said ChatGPT search uses third-party search providers as well as content provided directly by its partners.

That post is nearly two years old, and OpenAI has marked it as describing the 2024 launch. Check the current help page before you assume the setup hasn’t changed.

What does ChatGPT show?

Responses that use web search may include citations, and a Sources view lists cited and other relevant links. OpenAI is blunt about the limits: it says search results and citations can be incomplete, outdated or incorrect.

It advises users to open a cited source, check it supports the answer and review when it was published or updated. For you, that’s a hint worth acting on, because a visible date and a page that backs up its claims survive that scrutiny better.

What Are OpenAI’s Crawlers and What Does Each One Do?

OpenAI’s crawler page lists four user agents with different jobs, and the robots.txt settings for OAI-SearchBot and GPTBot are independent of each other. Mixing them up is an easy mistake, and it can cost you search visibility or give you less control over training than you intended.

User agent	What OpenAI says it does	Decides ChatGPT search appearance?
OAI-SearchBot	Surfaces websites in ChatGPT’s search features	Yes
GPTBot	Crawls content that may be used to train generative AI foundation models	No, it’s a separate setting
ChatGPT-User	Visits pages for certain user actions in ChatGPT and Custom GPTs	No, per OpenAI
OAI-AdsBot	Checks the safety of pages submitted as ads on ChatGPT	No, it only visits ad landing pages

OAI-SearchBot

OpenAI says sites opted out of OAI-SearchBot won’t be shown in ChatGPT search answers, though they can still appear as navigational links. It recommends allowing the bot in robots.txt and allowing requests from its published IP ranges.

If your site has allowed both OAI-SearchBot and GPTBot, OpenAI says it may use the results from a single crawl for both purposes.

GPTBot

GPTBot crawls content that may be used in training, and disallowing it indicates your content shouldn’t be used to train OpenAI’s foundation models. A site can allow OAI-SearchBot to appear in search while disallowing GPTBot.

Here’s what that split looks like in robots.txt. It’s an example of the documented setup, not a recommendation, because whether to block training is your call.

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

Related: Should You Block AI Crawlers Like GPTBot?

ChatGPT-User

OpenAI uses this agent for certain user actions, such as a user asking a question that sends ChatGPT to a page. It says the agent isn’t used to crawl the web automatically, that robots.txt rules may not apply because users initiate the requests, and that it isn’t used to decide whether content appears in search.

OAI-AdsBot

This one only matters if you advertise on ChatGPT. OpenAI says it visits pages submitted as ads to check they comply with its policies, and that the data it collects isn’t used to train foundation models.

How Do You Check That ChatGPT’s Crawler Can Reach Your Site?

Check three layers: your robots.txt, the CDN or firewall in front of your site, and your server logs. OpenAI documents the first two; the log check is our own routine.

Is robots.txt allowing it?

Open yourdomain.com/robots.txt and look for any rule that applies to OAI-SearchBot or to all user agents. A blanket “Disallow: /” under “User-agent: *” blocks it unless a more specific group allows it.

After you change the file, give OpenAI time. It says it can take around 24 hours from a robots.txt update for its systems to adjust for search results.

Is your CDN or firewall letting it through?

OpenAI says to confirm that your website host or content delivery network allows traffic from its published searchbot IP addresses. The ranges are listed in a JSON file at openai.com/searchbot.json.

This is the quiet failure. A bot-protection rule can refuse a crawler that your robots.txt welcomes, and nothing on the page will tell you.

Do your logs show it arriving?

Search your server or CDN logs for the user-agent text “OAI-SearchBot”. OpenAI’s example string includes a version number (1.4 at the time we read it), and it says the number may change, so match on the name rather than the version.

User-agent strings can be faked, so compare the source IP addresses with the published list, and look at the response codes. A run of 200s means the crawler is getting through; 403s mean something is refusing it.

A technical SEO audit usually picks up blocked bots and rogue rules in one pass. Keep in mind that a crawler visit proves access, not citation.

What if you want a page left out?

OpenAI says that if it obtains the URL of a disallowed page from a third-party search provider or by crawling other pages, and sees signals the page is relevant to a query, it may surface just the link and page title in ChatGPT Atlas. To prevent that, it recommends a noindex meta tag.

There’s a catch: the crawler must be allowed to crawl the page in order to read the tag. Blocking a page in robots.txt and adding noindex to it defeats the purpose.

How Do You Track Visits From ChatGPT?

Filter your analytics for utm_source=chatgpt.com. OpenAI’s publisher FAQ says publishers who allow OAI-SearchBot can track referral traffic from ChatGPT in platforms such as Google Analytics, because ChatGPT automatically includes that parameter in referral URLs.

What does the tag tell you?

It tells you which landing pages ChatGPT sends people to and how many visits arrive. That’s the only first-party measurement hook we found in OpenAI’s documentation, and there’s no Search Console equivalent for ChatGPT.

What does it miss?

Visits only show up if someone clicks a citation, and OpenAI says nothing about visits that arrive without the tag. A brand can be named in an answer without any click, so traffic understates visibility.

How do you measure the rest?

Run a fixed set of prompts on a schedule and record whether your brand is mentioned or cited. The method, including how many runs you need, is in our measurement guide.

Related: How Do You Measure AI Search Visibility?

What Makes a Page More Likely to Be Cited Once It’s Eligible?

OpenAI hasn’t said, so the useful answer is a set of reasoned bets: match the queries ChatGPT writes, be quotable, be checkable and be mentioned elsewhere. Everything in this section is inference from OpenAI’s description of the process plus outside research, and none of it is a documented ChatGPT requirement.

How do you match the queries ChatGPT writes?

If ChatGPT turns a prompt into narrower queries and reads what the search providers return, the pages that rank for those narrower queries are the ones it can use. So cover the follow-up questions a topic invites, not just the head term, as we explain for Google in the guide linked below.

Write pages that answer one question clearly, with the answer near the top. That’s good editorial practice rather than a documented ChatGPT rule.

Should you check Bing?

It’s a cheap precaution. OpenAI names Microsoft among its search providers, so confirm your key pages are indexed in Bing.

OpenAI hasn’t said how much weight any provider carries, so treat this as hygiene, not a lever.

How do you make a page quotable?

Publish original, specific material: first-party data, named methods and clear definitions give a model something to quote. That’s the work of content marketing when it’s done with a point of view.

The original GEO paper found that adding quotations, statistics and cited sources helped most in its tests. Its engines were a simulated one and Perplexity.ai, not ChatGPT, so it’s a lead rather than proof.

Related: What Is Generative Engine Optimisation (GEO)?

How do you make a page checkable?

Because OpenAI tells users to check publication dates and whether a source supports the answer, make both easy. Show a published or updated date, name your sources and keep claims specific enough to verify.

Do mentions on other sites help?

It’s reasonable to expect so. If ChatGPT retrieves pages through search queries, third-party pages that rank for your category’s questions can name you, and that is what digital PR earns.

This is our inference, not an OpenAI statement. Keep it genuine: coverage you’ve earned is worth having, and mentions you’ve manufactured are the sort of tactic Google’s guide calls unhelpful for its own AI features.

Want to see which of your pages ChatGPT can reach and what it says about your brand? Book an AI visibility audit, from £500.

What Doesn’t Work, or Isn’t Documented?

Treat anything OpenAI hasn’t documented as a theory, including llms.txt files, special markup and any ranked list of ChatGPT citation factors. If a guide gives you ten ranked factors, they were inferred from observation.

Do you need an llms.txt file?

The OpenAI crawler documentation and publisher FAQ we read don’t list llms.txt among the requirements. They name OAI-SearchBot access and IP access instead.

OpenAI’s own developer docs do publish an llms.txt index, but that tells us how OpenAI helps agents read its documentation, not whether ChatGPT search reads publishers’ files.

Related: What Is llms.txt and Does Your Site Need One?

Does keyword stuffing help?

In the GEO paper it offered little to no improvement, and on Perplexity.ai it performed 10% worse than the baseline. That doesn’t prove it hurts in ChatGPT, but there’s no evidence it helps.

Can anyone tell you the ranking factors?

OpenAI says only that ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information. It hasn’t published the factors or their weights.

What Don’t We Know About ChatGPT Citations?

Almost everything about why one page is cited over another. OpenAI says only that ChatGPT ranks search results using multiple factors and that placement is not guaranteed.

Which factors count, and by how much?

Unpublished. That includes how much any search partner’s index matters, and whether the same page is treated the same way in every country.

How stable are the answers?

Not very. SparkToro’s study with Gumshoe.ai had 600 volunteers run 12 prompts through ChatGPT, Claude and Google’s AI features, giving 2,961 runs in November and December 2025.

Rand Fishkin reported a less than 1 in 100 chance of the same list of brands appearing in any two ChatGPT or Google AI responses. Frequency was steadier than order: a cancer hospital appeared in 69 of 71 ChatGPT answers to one prompt, yet was the first mention in only 25.

Across 994 responses to headphone prompts that humans wrote, the top brands appeared 55 to 77% of the time. The study isn’t peer reviewed, and one of its authors works at an AI tracking vendor, which Fishkin discloses.

Do you see what your customers see?

Probably not. OpenAI says ChatGPT may use an approximate location from the IP address, that a VPN can change it, and that saved memories may shape the query when memory is on.

So a manual check from your own logged-in account is a sample of one, taken from an unusual position. That’s why a measurement routine uses a clean account and many runs.

How Long Does It Take to Show Up?

OpenAI documents one timing: around 24 hours for a robots.txt update to take effect for search. It publishes nothing on how long it takes to start being cited.

Step	What changes	Documented timing
Allow OAI-SearchBot in robots.txt	Your pages can be surfaced in search	Around 24 hours, per OpenAI
Fix CDN or firewall rules	The crawler can fetch your pages	None published
Publish or improve a page	A better candidate exists for retrieval	None published
Earn mentions elsewhere	Other pages can name you	None published

What should you expect in the first week?

Fix access, wait a day or two, then re-run your prompts. If nothing changes, that’s normal: access makes you eligible, and it doesn’t select you.

What is a sensible cadence?

This is our recommendation, not OpenAI’s. Log a baseline before you change anything, then re-run the same prompts monthly and again after any change you want to test.

Does the Same Approach Work in Other AI Engines?

Partly: the content advice carries over, but the access controls and reports are engine-specific. Each engine documents its own crawlers and its own eligibility rules.

Google AI Overviews and AI Mode

Google says a page must be indexed and eligible to show in Search with a snippet, and that there are no additional technical requirements. The controls and reporting differ from OpenAI’s, as the guide below sets out.

Related: How Do Google AI Overviews Choose Their Sources?

Microsoft Copilot

Microsoft’s Bing Webmaster Tools now reports citations for Copilot and Bing’s AI summaries. Microsoft doesn’t describe it as covering ChatGPT, so it can’t tell you what OpenAI does.

Perplexity, Claude and others

We haven’t read their publisher documentation for this guide, so we’re not making claims about them. Check each engine’s own pages, and treat third-party summaries with the caution you’d apply to any GEO advice.

Ready to see where ChatGPT stands on your brand today? Start with an AI visibility audit.