How to get recommended by Claude without guarantees
A practical Claude guide for clear sources, accessible pages, and separate evidence for answers, referrals, and ClaudeBot, Claude-User, and Claude-SearchBot requests.
No publisher can guarantee that Claude will recommend a brand. You can publish clear, inspectable information and check whether the relevant Anthropic bots can reach it, then measure answers and human visits separately.
Write for the decision behind the query
Explain what the product does, who it fits, where it is limited, what it costs or does not yet support, and which alternatives deserve consideration. A useful comparison gives the reader decision criteria instead of repeating a generic claim.
Make important claims inspectable
Use first-party documentation for product behavior and independent sources for market claims. Date facts that can change. Remove numbers you cannot reproduce, and link the method behind original research.
- Define the company and product in direct language.
- Keep feature, pricing, and legal copy consistent.
- State scope, dates, units, and limitations.
- Link to methodology and primary sources.
- Show counterexamples and cases where the product is not a fit.
Check the relevant Anthropic crawlers separately
Anthropic's site-owner guidance distinguishes three identities. ClaudeBot collects public web content that could contribute to model training. Claude-SearchBot accesses content for search results, while Claude-User supports user-directed fetching. Anthropic says its bots honor robots.txt and supports the non-standard Crawl-delay directive. A policy decision should be checked at the CDN, WAF, rate-limit, DNS, and origin layers too.
A request from any one of these identities proves only that the request occurred. It does not prove ingestion, training use, citation, recommendation, or a future answer.
Use the right measurement
When the answer is the subject, record an exact Claude query, date, location, model context, response, and links. When human behavior is the subject, use analytics for AI referrals, sessions, goals, and revenue. When access is the subject, use server evidence for bot identity, path, time, and HTTP outcome.
SearchSeal records the last category and provides privacy-friendly analytics for observed human activity. It does not convert a crawler request into a recommendation score.
A practical Claude review
Review the product definition, methodology, limitations, comparison criteria, and implementation documentation as separate pages. Then inspect the intended robots.txt policy and observed ClaudeBot, Claude-SearchBot, and Claude-User requests. If coverage is incomplete, report the unknown instead of concluding that Claude did or did not visit.
Specific, accessible, verifiable publishing may improve the evidence available to a system. It cannot guarantee that Claude will recommend the brand.