There's a particular irony in invisibility on Claude: it's often the result of a deliberate, poorly executed decision.

Since 2024, many companies have started blocking AI crawlers in their robots.txt to protect their training data. This is a legitimate decision. The problem: by blocking ClaudeBot — Anthropic's training crawler — many have accidentally blocked Claude-SearchBot at the same time, via generic rules. And Claude-SearchBot is the crawler that determines what Claude cites in its real-time responses.

According to ALM Corp's analysis of Anthropic documentation, this is "the most common self-inflicted GEO wound" observed in the market. Companies that have actively worked on their AI visibility — and made themselves invisible on Claude without knowing it.

The technical architecture you need to understand

Before analysing the causes of invisibility, you need to understand how Claude searches and cites sources — because it's fundamentally different from other AI engines.

Claude relies on Brave Search, not Google nor Bing

When Claude performs a web search, it uses Brave Search as its retrieval engine. This isn't a detail — it's the technical reality confirmed by TechCrunch (March 2025) and official Anthropic documentation.

The direct consequence: a Profound analysis from 2025 measured an 86.7% overlap between Claude citations and top Brave Search results. Your Google position isn't a reliable proxy for your Claude visibility. A page can be number 3 on Google and absent from Brave's top 20 — and therefore invisible to Claude during web searches.

Anthropic operates three distinct crawlers

This is the most poorly understood technical specificity of Claude:

  • ClaudeBot: collects data for future model training. Blocking it prevents your data from being used to train Claude, but doesn't directly affect your visibility in current responses.
  • Claude-SearchBot: indexing for real-time Claude search results. Blocking it makes you invisible in Claude responses.
  • Claude-User: real-time access during conversations, when a user asks Claude to analyse a specific URL. Blocking it makes your pages invisible during active evaluations by buyers.

The recommended configuration to maximise visibility while protecting your training data:

User-agent: ClaudeBot
Disallow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: Claude-User
Allow: /

The 6 real causes of invisibility in Claude

Cause 1 — You've blocked the wrong crawlers

This is cause number one — and the most immediately correctable. Check your robots.txt now. If you see any of these configurations:

User-agent: *
Disallow: /

# or

User-agent: anthropic-ai
Disallow: /

You're probably invisible to Claude-SearchBot and Claude-User. The fix takes 5 minutes and can have a visible impact in 2 to 4 weeks.

Cause 2 — Your site isn't well positioned on Brave Search

With 86.7% overlap between Claude and Brave Search, your Brave visibility is the technical prerequisite for your Claude visibility. Test now: go to search.brave.com and type your 3 main queries. If you don't appear in the top 10, Claude probably won't find you during web searches.

Submit your sitemap to Brave via search.brave.com/webmasters. It's free and takes 10 minutes.

Cause 3 — Your content doesn't pass Claude's verification filter

Claude is the most conservative AI engine in terms of citation. Its Constitutional AI approach pushes it to verify each important claim before citing it. Content that can't be independently verified — no cited sources, no dates on statistics, no identifiable author — has very little chance of being retained.

The Claude verification standard: each important statistic must have its source and date in the same sentence. Example of what Claude can verify: "According to Profound (2025), the overlap between Claude citations and Brave Search results reaches 86.7%." Example of what it cannot verify: "Studies show that AI visibility is growing strongly."

Cause 4 — Your author signals are absent

Claude applies a strict credibility hierarchy based on author signals. Observed data shows significant impacts on citation probability based on declared credentials: an author with 10+ years of declared experience in the field increases citation probability by 45%, recognised professional certifications by 60%.

An article without an identifiable author, without a bio, without a link to a verifiable professional profile is systematically disadvantaged by Claude — even if its content is excellent.

Cause 5 — Your entity isn't documented on multiple sources

Claude systematically cross-references multiple sources before citing a company. A company that only exists on its own site — without directory mentions, without presence on third-party platforms, without articles citing it — cannot be independently verified by Claude.

The practical rule: Claude needs to find your company on at least 2 to 3 independent sources to build a reliable entity. Your LinkedIn, a sector directory, and a press or analysis article mentioning you — that's the minimum base.

Cause 6 — Your content format isn't optimised for Claude extraction

Topify data on Claude citation patterns reveals a striking gap between formats: practical guides, tool pages and how-to tutorials obtain an average of 5+ citation appearances per page across 4 to 5 platforms. Standard blog articles obtain less than one citation. That's a gap of 30 to 50 times depending on format and depth.

The format that maximises Claude extraction: direct answer under each heading, followed by explanation, followed by a dated primary source. No long introduction delaying the answer.

The Claude paradox: quality required, but premium audience

Claude is the most demanding AI engine in terms of source quality. Being cited by Claude requires more work than being cited by ChatGPT or Gemini.

But the audience justifies the effort:

  • 80% of Claude users operate in a B2B context (versus 60% for ChatGPT)
  • 29% enterprise AI market share — ahead of OpenAI
  • More than 70% of Fortune 100 companies use Claude
  • Claude Pro subscribers have an average income 3.2 times higher than free ChatGPT users

A Claude citation in a B2B response carries more pipeline weight than a citation on almost any other AI platform. It's the most qualified audience — and the least contested in terms of optimisation.

Only 16% of brands systematically track their performance on AI platforms like Claude (Erlin, 2026). The remaining 84% don't know if they're visible or not.

The correction plan in order of priority

Priority 1 — Fix robots.txt (immediate impact, 5 minutes)

Check your robots.txt and ensure Claude-SearchBot and Claude-User are explicitly authorised. This is the fastest fix with the greatest potential impact.

Priority 2 — Activate on Brave Search (30 minutes)

Submit your sitemap on Brave Webmaster Tools. Test your 5 main queries on Brave and note your position. If you're not in the top 10, identify why — it's often an indexing or content structure problem.

Priority 3 — Make your content verifiable (2 to 4 weeks)

Reformat your main articles: identifiable author with bio and LinkedIn link, each statistic with source and date, links to primary sources, "direct answer → explanation → source" structure. This work has documented impact in 2 to 4 weeks on Claude citations.

Priority 4 — Build multi-source presence (2 to 6 months)

Identify the sources Claude considers authoritative in your sector — generally specialist publications, professional associations, B2B review platforms. Build documented presence on each. This is the longest but most durable lever.

Where to start?

Start by checking your robots.txt and Brave Search visibility — these two diagnostics take less than 15 minutes and can reveal immediate blockers.

Our free scoring tool tests your Claude visibility among the 5 analysed engines — with a structured score and first correction leads.

For a complete diagnostic with competitor analysis on Claude and precise identification of your blockers, our AI Diagnostic delivers a structured report within 5 business days.

To go further, also read How to check if your business appears in Claude and Why ChatGPT ignores your business.

Sources: ALM Corp analysis of Anthropic crawler documentation (2025-2026), Profound data on Claude-Brave Search overlap (86.7%, 2025), Topify data on content formats and Claude citations (2026), Erlin data on AI performance tracking (16% of brands, 2026).