Blog

Claude SEO: How to Get Cited by Claude

Claude SEO means making your brand citable when Claude answers buying questions: allow Anthropic’s crawlers (ClaudeBot, Claude-SearchBot, Claude-User), publish pages whose first sentences answer the question outright, keep your entity facts consistent everywhere Claude might read them, and earn mentions in the third-party sources it pulls into answers. Measured monthly, like everything in GEO.

Claude gets less GEO attention than ChatGPT, which is exactly why it is worth working on: the buying questions are increasingly asked there, and fewer competitors have done anything deliberate about it.

How Claude finds and cites brands

Like the other assistants, Claude blends two layers. Model knowledge, set during training, which you influence slowly by having a clear public footprint. And web search: for current or comparative questions Claude searches, reads a handful of pages and answers with citations. The second layer is where deliberate work shows up first, usually within 30 to 60 days.

In practice, Claude’s cited sources look familiar: comparison articles, documentation, review platforms, community discussions, and pages that answer the question directly. If you have already done serious work for ChatGPT visibility, most of it transfers. The differences worth knowing are below.

The crawler layer: three agents, three jobs

Anthropic uses distinct crawlers and your robots.txt should welcome all three. ClaudeBot collects for model training: blocking it keeps you out of future model knowledge. Claude-SearchBot indexes for search-backed answers: blocking it can cut how often you are cited. Claude-User fetches when a user asks about your page directly. To see whether your robots.txt lets each of them in, run your site through our free AI crawler checker.

Audit this now: many WordPress security plugins and CDN bot rules block these agents by default, and the site owner never chose that.

What Claude seems to reward

The public evidence base is thinner than for ChatGPT, so treat what follows as field observation from our own measurement rather than gospel, and re-test it against your own prompts.

Directness gets quoted. Claude’s answers favor sources that state things plainly and qualify them honestly. Pages that open with the answer, cite their own sources and avoid marketing inflation show up more consistently than louder alternatives.

Substance beats volume. A single deep, well-structured page on a topic outperforms five thin variants chasing keyword permutations. Consolidate; do not spray.

Precise claims with visible support. Numbers with sources, dated statements, named methods. Content that reads like a reference gets treated like one. When we tested which consulting firms Claude recommends, it warned that most “top AI consulting firms” lists “are written by firms that rank themselves,” and in several answers leaned on Forrester, Gartner and KLAS reports instead.

The work, in order

The Claude SEO checklist is the GEO checklist with the Anthropic specifics applied:

  1. Unblock the three crawlers and confirm your key pages render without JavaScript.
  2. Run the entity work so your name, description and facts are identical across site, schema, llms.txt and profiles.
  3. Rewrite your money pages answer-first.
  4. Earn mentions in the comparison lists and communities that show up when you ask Claude your own buying questions.
  5. Measure.

Measuring Claude specifically

Claude belongs as a distinct column in your measurement, not blended into an average: in the baselines we have run so far, our own site and our data study included, a brand’s share of voice on Claude often differs from its share on ChatGPT, in either direction. Fix the prompt set, run it monthly, score recommended, mentioned or absent, and count linked citations separately.

That per-engine discipline is the core of our Visibility Baseline. To see where you stand on Claude and the other four engines this week, the free AI Visibility Diagnosis covers it. Related: Gemini SEO, How to rank in ChatGPT and the GEO glossary.

If AI doesn’t name you, it names a competitor.

The free diagnosis runs real buying prompts on five engines, shows who gets named instead of you and where you can win first.