How to get cited by ChatGPT: 9 steps, from OpenAI's own docs

To get cited by ChatGPT, let OpenAI's search crawler in, make sure your pages are indexed by the search engines ChatGPT uses, write pages that answer the specific questions ChatGPT turns prompts into, state your facts in sentences it can quote, and get mentioned on the small set of sites it already trusts. OpenAI says placement is never guaranteed, but every step below comes from its own documentation or from measured citation data, not from guesswork.

I lead search and AI visibility at Fourward and built our free AI crawler checker. The nine steps are in the order I'd do them, with what each one fixes.

The nine steps at a glance

Fourward sells this work, so I have a stake in you caring about it. Everything about how ChatGPT behaves links to OpenAI's own pages, read on 23 October 2026. The citation shares come from Ahrefs' monthly study. Our own numbers come from a check we ran on 57 B2B SaaS sites on 21 October 2026.

How ChatGPT picks what to cite

Three things from OpenAI's documentation explain most of what follows.

• It searches with partners. OpenAI's help centre says ChatGPT search "sometimes partners with other search providers", rewrites your prompt into "one or more targeted queries", sends them, reviews the results, and may send "additional, more specific queries". The privacy policies it lists are Microsoft's and Shopify's.
• It has its own crawler for search. OAI-SearchBot "is used to surface websites in search results in ChatGPT's search features". Sites that opt out "will not be shown in ChatGPT search answers".
• It ranks, and nothing is guaranteed. The help centre says ChatGPT "ranks search results using multiple factors" and that "placement is not guaranteed". It also tells users that citations "can be incomplete, outdated, or incorrect".

Then there's who actually gets cited. Ahrefs tracks every domain ChatGPT cites across US queries. In its September 2026 update, Reddit took 16.8% of citations, Wikipedia 7.0%, Consumer Reports 3.7%, Forbes 3.1% and YouTube 2.5%. Ahrefs' own summary: ChatGPT "pulls from a relatively small set of trusted sources".

1. Allow OAI-SearchBot in robots.txt

This is the one step with a hard rule attached. OpenAI's crawler page says a site opted out of OAI-SearchBot won't appear in ChatGPT search answers, though it "can still appear as navigational links". Check your robots.txt for a block, then allow it explicitly:

User-agent: OAI-SearchBot
Allow: /

OpenAI says it can take "~24 hours" after a robots.txt change for its systems to adjust. Of the 57 B2B SaaS sites we checked in October, none fully blocked OAI-SearchBot, so if you do, you're behind every competitor on the list.

Here's what a block looks like. The New York Times disallows OAI-SearchBot in its robots.txt, along with ChatGPT-User, PerplexityBot and Claude-SearchBot. For a publisher licensing its content, that's a choice. For a SaaS company, it means ChatGPT can't cite you in search answers at all.

2. Let OpenAI's IP ranges through your firewall

Robots.txt is a request. Your CDN or firewall can still refuse the bot. The help centre's instruction for site owners is to allow OAI-SearchBot "and confirm that the website host or content delivery network allows traffic from OpenAI's published searchbot IP addresses", which it lists at openai.com/searchbot.json.

This is the step most sites miss. When we checked the agencies in our best B2B SaaS marketing agencies guide, every one allowed the bot in robots.txt, but three rejected at least one AI bot at the firewall. Our crawler checker tests both.

3. Get indexed by Bing

ChatGPT doesn't crawl the whole web itself for every answer. It rewrites your question and sends it to search partners, and the partner it names is Microsoft. If your page isn't in Bing's index, a rewritten query can't find it.

1. Verify your site in Bing Webmaster Tools. You can import a verified Google Search Console property in a minute.
2. Submit your sitemap and check the index coverage report for pages Bing skipped.
3. Use IndexNow for new and updated pages. Bing supports it, and most CMS platforms have a plugin.

4. Write for the rewritten query, not the prompt

OpenAI's own example: a researcher asks about drugs that target CCR8, and ChatGPT searches "CCR8 immunotherapy drug development 2025", then follows up with a more specific query. A buyer asking "what's a good CRM for a 20-person sales team" becomes a search for something like "best CRM small sales team 2026", then maybe "CRM pricing per user comparison".

So the pages that get found are the specific ones: a pricing page with the numbers on it, a comparison against the tool the buyer already knows, an integrations page per integration, a use-case page per role. Our GEO guide for B2B SaaS lists the page types buyers' prompts turn into.

5. State facts in sentences ChatGPT can quote

A citation attaches to a claim. "Plans start at $29 per user a month, billed annually" can be cited. A pricing table with no sentence around it is harder to quote, and a page that hedges everything gives the model nothing to point to.

• Who it's for, in one sentence, with the company size or role.
• What it costs, with the number, the unit and what's included.
• What it connects to, as a list of named products.
• What it doesn't do. Honest limits get cited too, and they build trust.

Put these near the top of the page, in plain text, not in an image or a tab that needs a click.

6. Get mentioned where ChatGPT already looks

This is the step that moves the needle most, and the one your own site can't do alone. Reddit takes 16.8% of ChatGPT's citations in Ahrefs' September data, Wikipedia 7.0%, YouTube 2.5%. For a B2B SaaS product, the practical list is:

1. Reddit threads where buyers ask for recommendations. Answer as yourself, say you work there, be useful. Spam gets removed and cited nowhere.
2. The comparison posts and "best X" lists that already rank for your category. Ask the authors to include you, with facts and a screenshot, not a pitch. Our pricing post exists partly because that's what people cite.
3. Review sites and directories your buyers use. Keep the listing current. An out-of-date price on a review site becomes an out-of-date price in an answer.
4. YouTube. A short, honest walkthrough with the product name in the title. It's the only video source in the top ten.

Ask ChatGPT your top 20 buyer questions and note every source it cites. Those sites are your outreach list.

7. Keep pages current and dated

OpenAI tells users to check "when it was published or updated" and to prefer current sources. Its own query example adds the year. Put a visible updated date on pages that change, update prices the day they change, and bump the date when you do. A pricing page last touched in 2024 loses to a competitor's touched this month.

8. Decide about GPTBot separately

OpenAI runs two crawlers for two jobs. OAI-SearchBot is for search. GPTBot crawls content "that may be used in training" its models. The docs say each robots.txt setting "is independent of the others", so you can block GPTBot and stay in ChatGPT search.

Whether to block training is a business decision. Of the 57 SaaS sites we checked, only Figma and Loom block GPTBot. What you shouldn't do is block GPTBot by copying a robots.txt template that also blocks OAI-SearchBot. Check both lines.

9. Measure whether it worked

1. Ask the questions monthly. The same 20 buyer prompts, in ChatGPT with search on. Open the Sources panel and record who gets cited.
2. Watch referrals. In Google Analytics, filter sessions from chatgpt.com. Clicks on citations show up there.
3. Track at scale. When you have more than a handful of prompts, an AI visibility tool runs them for you across ChatGPT, Perplexity and Google.

What doesn't help

• llms.txt. OpenAI's docs don't mention it, and Google says it doesn't use it. We looked at 57 SaaS companies' llms.txt files; it's harmless, but it isn't a step.
• Paying. ChatGPT ads sit below the answer and, OpenAI says, don't influence it. I covered what they do and don't do in ChatGPT ads for B2B SaaS.
• Blocking the bot "for now". Every day you're out of the index is a day the answer names someone else.

FAQs

Sources

1. Overview of OpenAI crawlers, OpenAI developer documentation.
2. Searching the web with ChatGPT, OpenAI Help Center, updated October 2026.
3. OAI-SearchBot published IP addresses, openai.com.
4. The 50 most-cited websites in ChatGPT (September 2026), Ahrefs.
5. Bing Webmaster Tools, Microsoft.
6. How Google interprets robots.txt, Google Search Central, for the llms.txt note.
7. The New York Times robots.txt, checked 23 October 2026.
8. Fourward AI crawler checker, used for the 57-site check on 21 October 2026 and the nytimes.com check on 23 October 2026.