Skip to content
Work
About
Blog
Talk to me

Engine guide: Claude

How to get cited by Claude

Claude looks at a different index from the Google and Bing based assistants. A site that is well covered by those two can still be thin or missing where Claude looks.

The short answer

When Claude searches the web, results come from Brave Search, which Anthropic lists as a web search subprocessor, and Brave runs its own independent index. Anthropic operates three agents: ClaudeBot for training, Claude-SearchBot for search quality, and Claude-User for fetches a person requests. To be cited, allow Claude-SearchBot and Claude-User, and check that Brave has indexed your pages.

What ranked on 26 September 2026: Anthropic's help centre page on its crawlers and trade coverage of the three-bot split rank here, next to tests comparing Claude's cited results with Brave's. Anthropic has not announced a search partner, so the Brave link rests on its subprocessor list and those tests.

What is specific to Claude

Two things set Claude apart: the index behind its search and the separation of its bots, each with its own robots.txt name.

  • Brave Search is built on its own crawl rather than on Google or Bing results, so coverage and ranking differ from both
  • ClaudeBot gathers training data; blocking it does not stop Claude-SearchBot or Claude-User
  • Claude-User fetches pages a user asks about and, per Anthropic, honours robots.txt, unlike some rival user agents
  • Web search is a feature Claude uses when a question needs current information, so many answers are not grounded at all
  • Citations appear inline next to the sentences they support, linked to the source page

Things that sound right and are not

  • ClaimA blanket AI-bot block is the safe default.

    Block lists copied from GitHub often include all three Anthropic agents. That removes you from Claude's search and fetches along with training. Choose per agent.

  • ClaimIf Google ranks us, Claude will find us.

    Brave's index is separate. Search your key queries on Brave directly; if you are absent there, that is the gap to fix before any content work.

  • ClaimClaude will cite our blog for everything.

    It cites the passage that supports a sentence in its answer. Pages that bury claims inside long narrative rarely get the inline link, even when retrieved.

Where to start

  1. 01Audit robots.txt and CDN rules for ClaudeBot, Claude-SearchBot and Claude-User, and set each one on purpose.
  2. 02Search your top ten buyer questions on search.brave.com and note whether your pages appear.
  3. 03If important pages are missing, check they are linked from pages Brave does index, since Brave discovers mostly through links and its own crawl.
  4. 04Put supporting facts in short, attributable sentences that could sit next to a citation marker.
  5. 05Test the same questions in Claude with web search on, and record which URLs are linked.

Questions

Can I submit my site to Brave?
Brave has offered a page for suggesting URLs, but inclusion depends on its crawler. Solid internal and external links remain the dependable route.
Does blocking ClaudeBot affect citations?
Anthropic states the three agents are controlled separately, so blocking ClaudeBot alone should not stop search or user fetches.
Does Claude use Google?
Anthropic has not said so. The documented vendor on its subprocessor list is Brave, and independent tests match Claude's citations closely to Brave results.

Related

Doing this yourself and want a second opinion on what the assistants say about you? Tell me what you are seeing, or read the AEO Playbook for the whole framework.