Claude now uses Brave Search for web browsing, in addition to training data. Your content needs to be deeply factual, well-structured, and indexed in Brave's search engine. Google or Bing indexing is not enough. Bullet-pointed pages are 30% more likely to be cited.

How Claude differs from every other AI platform

Claude combines training data with real-time web browsing powered by Brave Search. This makes it different from ChatGPT (which uses Bing) and Gemini (which uses Google). Key implications:

Brave Search for browsing - Claude uses Brave's index for real-time web lookups, not Google or Bing
Training data as baseline - authoritative pages absorbed during training form Claude's core knowledge
Brave indexing required - browse your site in Brave browser with data sharing enabled and visit 20+ pages. Google or Bing indexing does NOT guarantee Brave indexing
Depth over recency - comprehensive pages are favoured over recently published ones
Less volatile citations - once Claude cites a source, it tends to keep citing it. ChatGPT shifts more frequently

Claude's market position

Metric	Value	Context
Market share	2.91%	Doubled in a single month
Search engine	Brave Search	Uses Brave's index for real-time browsing (not Google or Bing)
Crawl-to-referral ratio	23,951:1	Crawls heavily, refers very little traffic
Growth trajectory	Doubling monthly	Fastest-growing AI platform by percentage
Citation volatility	Low	More stable than ChatGPT or Perplexity

Claude's crawl-to-referral ratio of 23,951:1 means it crawls your content heavily but sends almost no referral traffic. Perplexity is the opposite, with the best ratio at under 200:1. Claude's value is brand visibility within answers, not click-throughs.

What drives Claude citations

1. Structured content with bullets and lists

Pages with bullet-pointed content are 30% more likely to be cited by Claude. This makes intuitive sense - structured content is easier for language models to parse and extract factual claims from during training.

2. Factual density

Claude rewards pages that pack specific data points, statistics, and verifiable claims into concise formats. Vague marketing copy gets ignored. Specific numbers get absorbed.

3. Comprehensive coverage

Depth matters more than breadth for Claude. A single page that thoroughly covers a topic outperforms ten shallow pages. Claude's training process favours authoritative, complete resources over fragments.

4. Authoritative source domains

Content on recognised, high-trust domains has a higher chance of being included in training data. Academic papers, industry publications, established media, and well-known brand websites all carry more weight.

5. Clean, parseable HTML

Like all AI crawlers, ClaudeBot reads HTML source. Heavy JavaScript rendering, pop-ups, and dynamic content loading create barriers. Static HTML with clear heading hierarchy is ideal.

What to optimise for Claude specifically

Action	Priority	Why it matters for Claude
Use bullet points and numbered lists	High	30% citation increase for structured content
Include specific statistics	High	Factual density drives training absorption
Write comprehensive, long-form content	High	Depth over breadth for training selection
Publish on authoritative domains	High	Domain trust affects training inclusion
Use clean HTML (avoid heavy JS)	Medium	ClaudeBot crawls HTML source
Allow ClaudeBot in robots.txt	Essential	Blocked = never crawled = never trained on
Implement schema markup	Medium	Adds structured signals during crawl

Robots.txt configuration for Claude

Anthropic operates multiple crawlers. ClaudeBot collects training data. Claude-SearchBot and Claude-User handle real-time search and user-initiated browsing via Brave Search. Add all three to your robots.txt:

User-agent: ClaudeBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: Claude-User
Allow: /

User-agent: anthropic-ai
Allow: /

Check your current configuration. Many WordPress security plugins block all bots by default, including ClaudeBot. Also ensure your site is indexed in Brave Search. Browse your own site in Brave browser with data sharing enabled. Visit 20+ pages. Google or Bing indexing does NOT guarantee Brave indexing.

The crawl-to-referral ratio problem

Claude's 23,951:1 ratio is the worst among major AI platforms. For comparison:

Perplexity: under 200:1 (best)
ChatGPT: approximately 1,000:1
Gemini: approximately 3,500:1
Claude: 23,951:1 (worst)

Optimising for Claude is about brand visibility within answers, not driving traffic. When Claude mentions your brand in a response, users see it. But they rarely click through. The value is in awareness and authority.

Why Claude still matters

Despite the low referral ratio, Claude matters for three reasons:

Fastest growth - 2.91% market share that doubled in one month means it could be a major player within a year
Professional audience - Claude users skew toward developers, researchers, and business professionals
Stable citations - once you earn a Claude citation, it tends to persist across conversations, unlike the volatility seen in ChatGPT

How to Get Cited in Claude - What We Know