How do I get cited by ChatGPT?

Publish a crawlable page that answers one specific question directly, allow OAI-SearchBot and ChatGPT-User, make the page visible in Bing, and verify in server logs that search and AI retrieval agents can fetch it without 403, challenge pages, or broken TLS.

Can llms.txt guarantee ChatGPT citations?

No. llms.txt can point assistants to preferred source pages, but citation likelihood depends more on crawl access, indexing, answer quality, and whether the page is useful for the user's query.

Question

How to get cited by ChatGPT

Updated June 8, 2026

To get cited by ChatGPT, publish a page that can be crawled, indexed, understood quickly, and quoted without guesswork. There is no form that guarantees a citation. The practical work is to remove crawl friction, answer a specific question better than competing pages, and verify that OpenAI and search crawlers can fetch the URL.

Step 1: Let the right bots in

Start with the access layer. A useful answer page is irrelevant if the crawler sees a WAF challenge, 403, or broken TLS.

User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

Step 2: Make one quotable answer per page

ChatGPT citations are most likely when the page contains a direct answer to the exact query. Do not bury the answer behind a long introduction.

Put the literal question in the title and H1.
Answer it in the first one or two sentences.
Use dates where the answer is time-sensitive.
Back the answer with a table, ordered checklist, command block, or clear comparison.
Link to official documentation or primary evidence when the claim depends on a vendor policy.

Step 3: Be machine-readable

Serve static HTML; do not require JavaScript to see the content.
Add FAQPage JSON-LD whose answer text matches the visible answer.
Use a canonical URL and avoid duplicate versions on apex/www or tracking parameters.
Keep code samples and configuration examples in plain text, not screenshots.

Step 4: Reach Bing and Google

Do not rely on passive discovery for a new site. Submit updated URLs through IndexNow for Bing-side discovery, keep XML sitemaps current, and use Google Search Console URL Inspection for priority pages. If the URL is not indexed anywhere, it is much less likely to be selected as a citation source.

Step 5: Build evidence into the page

A page that merely says "best practice" is weak. A page that shows commands, exact headers, error codes, example output, and decision boundaries is easier to cite. For technical pages, include the shortest reliable diagnostic path:

# Example: confirm that search/retrieval bots can reach the page
curl -I https://example.com/your-page/
curl -A 'OAI-SearchBot' -I https://example.com/your-page/

Step 6: Verify access in logs

grep -Ei 'OAI-SearchBot|ChatGPT-User' /var/log/your-access.log | awk '{print $7}' | sort | uniq -c | sort -nr | head

Separate crawler access from user traffic. OAI-SearchBot requests suggest search discovery. ChatGPT-User requests may indicate live user retrieval. Either way, look for 200 responses on the pages you want cited.

Common reasons ChatGPT does not cite a page

The page is not indexed or discoverable for the query.
The CDN blocks unfamiliar user agents or non-browser requests.
The answer is too generic compared with stronger competing pages.
The useful detail is hidden behind JavaScript, tabs, images, or login.
The page title targets one query, but the body answers a different question.

Use the GEO checklist