Engine visibility guide

ClaudeBot traffic and Claude Search visibility are different signals.

Anthropic documents separate agents for potential training collection, search indexing, and user-directed retrieval. A useful Claude scorecard identifies the agent or answer surface first, then reports what the retained answer actually did with the brand.

Does a ClaudeBot request mean Claude Search can see or cite us?

No. Anthropic identifies ClaudeBot as its potential-training crawler, not its search-indexing agent. Search exposure should be evaluated with verified Claude-SearchBot activity, Claude-User retrieval where applicable, and completed Claude answers in which web search was actually enabled or invoked. Even those crawler events establish only access or retrieval; they do not guarantee that Claude will mention, cite, or recommend the brand. Ready Scan therefore records the exact buyer question, model and surface, search state, answer, citations, recommendation status, competitors, locale, timestamp, and limitations for each Claude observation.

Claude’s three-agent decision map

The three documented Anthropic agents belong in different telemetry buckets. Keeping those purposes separate prevents a large training-crawler number from being marketed as Claude Search visibility.

  1. ClaudeBot

    Collects public-web content that could potentially contribute to model training. A fetch does not prove training inclusion or search selection.

  2. Claude-SearchBot

    Supports search indexing and result quality. Anthropic says blocking it may reduce search visibility, which is an eligibility statement rather than a placement guarantee.

  3. Claude-User

    Retrieves pages in response to user-directed activity. Its request is evidence of that retrieval path, not a general index or endorsement.

Run a reproducible Claude visibility panel

Use stable category, comparison, use-case, and trust questions with web search explicitly enabled. Preserve whether search ran, any generated query, the cited URL and title, the cited answer text, brand treatment, recommendation strength, competing domains, model, locale, and time. Anthropic’s API exposes structured web-search results and citations, but controlled API results must remain labeled separately from consumer Claude observations.

Apply this to your answer market

Check your brand in web-search-enabled Claude answers.

Keep Claude Search citations and recommendations separate from training-crawler and user-retrieval traffic.

Check my brand
Continue the strategy