Citino teamHow Claude answers
Claude answers in two ways. Without web search it answers from what the model learned in training, which changes only when a new model ships. With web search, which Anthropic launched in March 2025 and made available on all plans that May, it retrieves pages and cites them.
Which search engine powers it is not stated on Anthropic’s announcement. In March 2025 TechCrunch reported that Anthropic had added Brave Search to its subprocessor list. If that still holds, your visibility in Brave Search matters for Claude, and it is worth checking your key pages at search.brave.com.
Allow the right Anthropic crawlers
From Anthropic’s crawler documentation:
| Bot | What it does | Block it and… |
|---|---|---|
Claude-SearchBot | Crawls to improve search result quality | Your content is not indexed for Claude’s search |
Claude-User | Fetches pages when a user asks Claude a question | Your visibility in answers to user queries can drop |
ClaudeBot | Collects content for training | Your future content is kept out of training |
Anthropic says its bots honour robots.txt, including the non-standard Crawl-delay, and that you need a rule for each subdomain. It also warns that blocking by IP address may not work correctly; use robots.txt.
To be recommended while opting out of training:
User-agent: Claude-SearchBot
Allow: /
User-agent: Claude-User
Allow: /
User-agent: ClaudeBot
Disallow: /
What makes Claude cite a source
Claude tends to be careful about what it cites, and it favours sources that are precise. In practice that rewards:
- Specific, checkable facts. Exact prices, named integrations, dated changes. Vague claims give it nothing to quote.
- Sources for your claims. Link the evidence: documentation, public reports, standards.
- Independent confirmation. Self-ranked “best of” lists carry little weight. Reviews, independent comparisons and practitioner discussion carry more.
- Readable pages. Text in the HTML, clear headings, tables for comparisons.
How to check where you stand
Ask your buyers’ questions in Claude with web search on, several times each, in a fresh conversation. Record whether you are named, your position, the tone, and the sources cited. As with every engine, compare rates over many runs, not single answers.
How Citino tracks Claude
Citino measures Claude as an add-on to any plan, alongside the three included engines (ChatGPT, Google AI Overviews and Perplexity), with every answer kept and every number carrying a margin of error.