Guide

How to get cited by Claude

Published 1 July 2026 · 8 min read

Claude cites a source when its search crawler can reach the page, the answer is in the page's text, and a passage cleanly resolves the question. To get cited, allow Claude-SearchBot, lead with your answer, and keep the page current — here's how each part works.

How Claude finds and cites sources

Claude answers a question with web search by reading a handful of pages and naming the ones it drew from. Anthropic runs three separate crawlers for this: Claude-SearchBot indexes pages for Claude's search, Claude-User fetches a page live when someone asks Claude about it, and ClaudeBot collects training data. A citation depends on the search crawler reaching a page that answers cleanly — not on ranking in a list.

Step 1: Let Claude-SearchBot reach your pages

Access is the foundation. Check yoursite.com/robots.txt and confirm Claude-SearchBot isn't disallowed, including by a catch-all User-agent: * with Disallow: / that blocks every unnamed crawler. Anthropic states that blocking Claude-SearchBot prevents your content from being indexed for Claude's search results, which removes you from the answers it cites.

# Allow Claude's search & retrieval crawlers User-agent: Claude-SearchBot User-agent: Claude-User Allow: /

The bot people block by mistake: ClaudeBot

ClaudeBot is the crawler most often confused with the others. ClaudeBot collects content for Anthropic's model training, not for search — so blocking ClaudeBot opts you out of training but does not remove you from Claude's cited answers. The crawler you must keep unblocked for citations is Claude-SearchBot, not ClaudeBot.

Step 2: Put the answer in the first sentence

Claude rewards pages that resolve the question early. A passage that answers in its opening sentence, under a heading that matches the question, is far easier for Claude to lift and attribute than one buried beneath several paragraphs of build-up. Write each section so its first line is the takeaway.

Step 3: Make each passage self-contained

Claude lifts a chunk, not a whole page. A paragraph that only makes sense after the one before it is hard to quote cleanly, so avoid opening with "It," "This," or "That," and name your entities plainly — "robots.txt," "Claude-SearchBot" — rather than "the file" or "the crawler." Our guide on making pages quotable covers the structure in detail.

Step 4: Keep the page current

Freshness helps for anything time-sensitive. A maintained, clearly dated page is a stronger candidate than one that looks abandoned, so revisit your key pages rather than publishing once and leaving them. Accurate, up-to-date content is part of staying citable.

Claude compared with other engines

The same reach-read-quote logic applies across answer engines. Like OpenAI's OAI-SearchBot for ChatGPT, Claude-SearchBot is a search crawler whose access decides citation eligibility, and the structural work that helps one engine helps the others. Our GEO playbook covers the shared foundation, so a page built for Claude is usually well-placed for ChatGPT, Perplexity and Google's AI Overviews too.

Common reasons Claude doesn't cite a page

Most misses trace back to a short list. Claude-SearchBot may be blocked in robots.txt; the content may load with JavaScript and arrive as an empty shell; the answer may be buried where Claude can't lift it; the page may be stale on a question where recency matters; or the content may be thin and unsourced next to stronger competitors. Each one is fixable once you know which applies.

A quick before-and-after

A short rewrite shows the difference. A page that opens "There are several things that can affect whether AI assistants mention you, and it varies" gives Claude nothing to lift. Rewritten to "Claude cites your page when Claude-SearchBot can crawl it and a sentence answers the question directly," the same page now leads with a self-contained answer Claude can quote and attribute. The fix is specificity, not length.

How to check whether Claude cites you

Test it two ways. Ask Claude (with web search enabled) a question your page should answer and see whether it cites you — treat the result as a hint, since it varies by wording. Then check your server logs for Claude-SearchBot and Claude-User visits. The free Site Check at Sailiently confirms whether the search crawler can reach and read your page, with the evidence behind each result.

What you can't force

One honest limit applies here too: nothing guarantees a Claude citation. The steps above make your page reachable, readable and quotable — a strong candidate — but whether Claude quotes you on a given question depends on the query and the competing sources. Be wary of anyone promising guaranteed citations.

Does Claude train its models on my content?
ClaudeBot collects content for Anthropic's model training, and you can opt out by disallowing it in robots.txt. Claude-SearchBot is separate — it indexes pages for Claude's search results, not for training.
What's the difference between Claude-SearchBot, Claude-User and ClaudeBot?
Claude-SearchBot indexes pages for Claude's search; Claude-User fetches a page live when someone asks Claude about it; ClaudeBot collects training data. Each is controlled separately in robots.txt.
Will blocking ClaudeBot remove me from Claude's answers?
No. ClaudeBot is the training crawler. Claude's cited answers come through Claude-SearchBot, so blocking ClaudeBot opts you out of training without affecting whether Claude cites you.
Should I block Claude using IP rules?
Anthropic advises against IP-based blocking, because it can stop the crawler from reading your robots.txt and won't reliably persist. Use robots.txt directives to control Claude's crawlers instead.
How quickly does Claude pick up my changes?
There's no fixed schedule, and it varies with your site's freshness and importance. After a robots.txt or content change, allow a few days and re-check rather than expecting an instant result.
Is getting cited by Claude different from ChatGPT or Perplexity?
The logic is the same — reachable, readable, quotable — only the crawler names differ. A page built to be cited by Claude is usually well-placed for ChatGPT and Perplexity too.

Sources

The crawler facts above come from each engine’s own documentation:

See where your site stands — free.

Run a Site Check on any page in under a minute. No card, 15 free credits to go deeper.

Check your site →

Related: How to get cited by Perplexity · Is your site readable by AI crawlers?