How to get cited by ChatGPT

How to get cited by ChatGPT is mostly a fetch-and-extract problem. Allow OAI-SearchBot, write the answer first, and match buyer questions. OpenAI documents that sites opted out of OAI-SearchBot will not appear in ChatGPT search answers. Put a self-contained answer in the first paragraph as plain text. Blocking GPTBot only opts you out of training, not of search citations.

How to get cited by ChatGPT?

Do the three things OpenAI actually documents, in this order. Skip the rest until these pass.

• Be fetchable by the search crawler. Allow OAI-SearchBot in robots.txt, and allow the published IP ranges through any CDN or WAF that filters bots. OpenAI's crawler docs say sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links.
• Put the answer in text, up front. ChatGPT search cites pages it can retrieve and quote. A hero image, a client-rendered app, or a preamble that announces the article gives it nothing to lift. Write the answer in the first paragraph, in 40 to 90 words, as a complete sentence a model can extract.
• Match a real question. Citation happens when a retrieved passage answers the prompt. Question-style headings, a visible FAQ, and FAQPage JSON-LD that matches the on-page text make that passage easy to find. Google's structured data policies require the markup to match the visible page. The same rule is the honest one for any consumer of the JSON-LD, including answer engines.

There is no ChatGPT-only file format, no paid inclusion, and no ranking factor you can buy. OpenAI's help article is blunt: ranking is based on factors meant to surface reliable, relevant information, and there is no way to guarantee top placement.

A working robots.txt split looks like this:

User-agent: OAI-SearchBot
Allow: /
User-agent: ChatGPT-User
Allow: /
User-agent: GPTBot
Disallow: /

Confirm the live file, not the file you think you deployed. Our AI crawler checker shows the exact rule that decided each verdict. The longer writeup is AI crawlers and robots.txt.

Which OpenAI crawlers decide whether you appear?

OpenAI runs separate user-agents for separate jobs. Treating them as one "ChatGPT bot" is how people opt out of citations by accident.

• OAI-SearchBot builds the index behind ChatGPT search. This is the one that decides whether you can be shown as a cited source in search answers. Allow it if citations are the goal.
• GPTBot crawls for training data. Disallowing GPTBot tells OpenAI not to use the site to train foundation models. It does not, by itself, remove you from ChatGPT search.
• ChatGPT-User fetches a page because a person asked about it, or because a Custom GPT action requested it. OpenAI says it is not used for automatic crawl, is not used to decide Search inclusion, and that robots.txt rules may not apply because the request is user-initiated.

The practical split: allow OAI-SearchBot (and usually ChatGPT-User), then decide on GPTBot as a licensing call. OpenAI also notes the two allowed crawls may share a fetch so they do not hit you twice.

ChatGPT search sometimes partners with other search providers. Classic SEO still matters, because a partner index will prefer pages that already rank for the rewritten query. That is not a reason to ignore OAI-SearchBot. It is a reason not to treat ChatGPT as a separate web.

What page structure does ChatGPT actually quote?

A citation is a quoted span plus a link. Pages that get cited tend to have a span worth quoting.

Write the answer first. A self-contained opening paragraph that names the topic and states the fact is what an answer engine can lift. Forty to ninety words is the band our content spec uses, because it is long enough to be complete and short enough to extract. A quotability pass on the AEO checker will tell you whether that opener exists.

Then map the rest of the page onto questions:

• Question headings, not topic labels. "How to get cited by ChatGPT?" survives extraction. "ChatGPT citation tips" often does not.
• Short, visible FAQ answers. The same questions belong in FAQPage JSON-LD, matching the visible text.
• Article JSON-LD for headline, description, author and datePublished, so the engine does not have to infer them.
• Lists and definitions for steps and splits (the three crawlers, the three checks). Models quote those cleanly.

FAQ rich results in Google Search are no longer a reason to add this markup for most sites. Google limited FAQ rich results to well-known government and health sites in August 2023. Ship the JSON-LD anyway: answer engines and other consumers read it as facts, not as a snippet enhancement.

llms.txt does not substitute for any of this. OpenAI's crawler documentation never mentions it. Some coding agents fetch it when it exists, which is why we still publish one. It is a briefing, not a citation switch. See what llms.txt actually does.

How is a citation different from a brand mention?

A mention is your name in the prose. A citation is a clickable source. You can have one without the other, and they fail for different reasons.

• Mention without citation: the model knows the brand (from training, from a partner result, or from a page it read and did not link). Useful for awareness. No traffic.
• Citation without a flattering mention: you are a footnote. Traffic is possible, the recommendation went to someone else.
• Neither: you are not in the retrieved set. Start with crawler access and extractable pages before you rewrite the homepage copy.

Do not test this with a prompt that names your brand. "What is HuskyScout?" will mention HuskyScout. The prompts that reveal citation are the buyer questions where you have to earn the link: "how to get cited by ChatGPT", "best AI visibility tools for a two-person SaaS", "which crawlers does ChatGPT use". Keep that set fixed so the next run is comparable. The method is in how to check whether AI answers mention your brand.

How do you check whether ChatGPT is citing you?

Ask the buyer questions, with search enabled, and record the cited URLs. One screenshot is not a measurement. Answers vary between runs.

For each prompt, store:

• Whether your brand was named.
• Whether your domain appears in the citation list (the inline marks or the Sources panel OpenAI documents in the ChatGPT search help).
• Which other domains were cited instead. That list is the work queue.

Re-run weekly. OpenAI says search systems can take about 24 hours to adjust to a robots.txt change. That is the policy recrawl, not a full reindex of every page. Content and ranking movement takes longer. Daily checks mostly measure answer variance.

If citations are missing, walk the floor in order: crawler allowed, IP ranges not blocked, answer in the first paragraph, question headings present, then the prompt set. Our AI visibility checker and the full probe run that matrix so you are not doing it by hand.

Frequently asked questions

Can I block GPTBot and still get cited by ChatGPT?

Yes. OpenAI's crawler documentation treats GPTBot and OAI-SearchBot as independent. Disallow GPTBot to opt out of training. Allow OAI-SearchBot if you want ChatGPT search answers to include you. After you change robots.txt, fetch the live file, because a wildcard Disallow can still catch the search bot.

Does publishing llms.txt get me cited by ChatGPT?

No published OpenAI document says ChatGPT search reads llms.txt. Google's AI-features guidance is explicit that you do not need new machine-readable files to appear in AI Overviews, and that is a different product. llms.txt still helps agents that choose to read it. It is not a citation lever for ChatGPT search.

How long until a robots.txt change shows up in ChatGPT search?

According to OpenAI's crawler documentation, it can take about 24 hours from a robots.txt update for search systems to adjust. That is the recrawl of the policy, not a promise that every page is reindexed. Re-run the same buyer prompts a week later. Daily checks mostly measure answer variance.

If Google AI Overviews cites me, will ChatGPT cite me too?

Not automatically. Google AI Overviews use Googlebot and the same technical requirements as Search, according to Google Search Central. ChatGPT search uses OAI-SearchBot and may also query partner search providers. Being eligible for one index does not enroll you in the other.