How to get cited by ChatGPT search: a technical checklist

1. Let OAI-SearchBot reach the page

ChatGPT search relies on OpenAI's search crawler, OAI-SearchBot. OpenAI recommends allowing it in robots.txt and allowing requests from its published IP ranges. Sites that opt out are not shown in ChatGPT search answers, although they can still appear as navigational links.

• Open https://your-site.com/robots.txt and confirm that no group blocks OAI-SearchBot, including a broad User-agent: * rule with Disallow: /.
• Check your CDN or firewall for AI crawler blocks and bot challenges. A robots.txt allowance does not help if the request is stopped before it reaches your server.
• Blocking GPTBot is fine if you do not want training use. It is a separate crawler and does not affect search. See GPTBot vs OAI-SearchBot.

2. Make the page indexable and readable as HTML

An answer engine can only quote text it can fetch and read.

• The canonical URL returns HTTP 200, not a redirect chain or a soft error page.
• There is no noindex in a <meta name="robots"> tag or an X-Robots-Tag header on pages you want cited.
• The canonical tag points to the public, preferred URL.
• The main text is present in the initial HTML. Not every crawler renders JavaScript the way a browser does, so a page that shows its content only after scripts run is a risk. Compare the output of curl -s https://your-site.com/page with what you see in the browser.
• An XML sitemap lists the pages and is referenced in robots.txt.

Our 15-minute visibility check shows how to test each item with ordinary tools.

3. Answer the question in the first paragraph

Answer engines assemble responses from passages. A passage that answers the question directly is easier to use than a paragraph of brand language.

• Open with a concise answer of roughly 40 to 120 words: what the thing is, who it is for, and the key fact or number.
• Use headings that match the questions people ask, such as "How much does it cost?" or "Does it work with Shopify?".
• Put comparisons in tables and procedures in numbered steps.
• Name things precisely. "Our platform" is vague; "the generator writes robots.txt rules for 11 named crawlers" can be quoted and checked.

4. Make every important fact verifiable

A page that shows its evidence is a safer source to cite.

• Show who wrote or reviewed the page and when it was last updated.
• Link to primary sources for claims, prices and statistics, and keep those numbers current.
• Use structured data that matches the visible content: Organization with consistent name, logo and profiles, Product or SoftwareApplication for what you sell, Article for editorial pages. Never mark up reviews, ratings or awards that are not real.
• Keep the same facts consistent across your site, your social profiles and directory listings.

5. Be findable beyond your own site

Answer engines draw on many sources, not only the pages you publish. Accurate descriptions of your business on reputable third-party sites help, for example industry directories, review platforms, partner pages, documentation and press coverage. Earn those mentions by being useful. Paid link schemes, fake reviews and automated forum posts can damage trust and may break platform rules.

6. Measure with a fixed prompt set

You cannot improve what you do not measure consistently.

1. Write 10 to 20 prompts that real customers would ask, such as "best [category] for [audience]" or "how do I [task] with [tool]".
2. Run the same prompts on a regular schedule and record the full answer, whether your brand is mentioned, and which URLs are cited.
3. Compare runs over several weeks before calling a trend. Answers vary with wording, location, personalization and model updates.
4. In your analytics, look for referral visits from chatgpt.com. Many ChatGPT links carry utm_source=chatgpt.com, but visits from apps or copied links can arrive as direct traffic, so treat the count as a minimum.

What does not work

• Hidden text or instructions aimed at AI crawlers. This is manipulation, and it can backfire badly if discovered.
• Mass-produced pages that target every possible prompt without adding real information.
• Buying or faking reviews and endorsements.
• Expecting llms.txt to act as a switch. We cover the evidence in Does llms.txt work?.

Sources

• OpenAI: Overview of OpenAI crawlers
• Google Search Central: AI features and your website
• Google Search Central: Creating helpful, reliable, people-first content

Reviewed on 2026-10-06. This checklist describes technical and editorial readiness. It is not a prediction that any engine will cite a particular page.