How to Rank in Claude: Get Your Site Cited by Anthropic's AI in 2026

Quick quiz

How should you get mentioned by ChatGPT and other AI assistants?

Question 1 of 425%

What do you want to know or change?

Just see if AI assistants talk about my businessFollow it over time against my competitorsBe mentioned more often, starting now

Ranking in Claude means being one of the sources Claude names when someone asks a question about your market. Anthropic launched web search in Claude on March 20, 2025, and made it available globally on all plans on May 27, 2025. Since then, Claude can read the web in real time and gives direct citations so users can check what it says. This guide completes our series on ranking in ChatGPT, Perplexity and Gemini with what is specific to Anthropic's assistant.

How does Claude find and cite sources?

Claude has two ways of knowing about your brand, and they call for different work.

1. Training memory: what the model learned from public web content before its cutoff date. It shapes how Claude describes a brand without searching, and it changes slowly, at each new model.
2. Web search: when a question needs recent or precise information, Claude decides to search, turns the question into one or several queries, reads the most promising pages and writes an answer with inline citations.

For web search, Anthropic relies on an external index. In March 2025, TechCrunch reported that Anthropic had added Brave Search to its list of subprocessors, and that Claude's citations matched Brave's results on test queries. Anthropic has not published how it ranks results, so treat Brave Search as the most likely entry point, not as the only one. The practical consequence is clear: a page Brave does not know is unlikely to be found by Claude's search.

The last column comes from our Claude citation study, measured on 20,011 Claude answers between August 2025 and August 2026. Claude names few sources, which makes each citation harder to win and more valuable once won.

Which Anthropic crawlers should you allow?

Anthropic documents three bots in its help center article on crawling, and each one has a different role. Blocking one does not block the others.

To be cited, Claude-SearchBot and Claude-User must stay allowed. ClaudeBot is a business choice: blocking it keeps your new content out of training, at the cost of a weaker presence in what future models know without searching. Anthropic honors robots.txt, supports the non-standard Crawl-delay directive, and publishes the IP addresses of its crawlers so you can verify that a visit is genuine. Our guide to robots.txt for AI crawlers gives ready-to-use rules for every assistant, and the AI crawlers entry of our glossary lists the other agents.

What kind of sources does Claude cite?

Our panel of buyer questions, sent every month to five AI assistants, shows that Claude is the most selective of them. Across 20,011 measured answers, its most cited platform is Trustpilot (1.57% of answers), followed by HubSpot (0.5%), Amazon (0.49%), Indeed (0.28%) and LinkedIn (0.25%). The top three platforms hold 63% of the citations counted in the study's ranking.

Two lessons follow. First, apart from Trustpilot, no platform reaches 1% of Claude's answers, so most of its citations go to a long tail of specialized and brand sites: your own pages must be the clearest answer available. Second, the platforms it does use repeatedly are review and reputation sites, where buyers compare suppliers. Our Trustpilot study details how often each assistant relies on it.

How to rank in Claude, step by step

1. Open access to the right bots

Allow Claude-SearchBot and Claude-User in robots.txt, check that your CDN lets them through, and serve your main content in the HTML rather than only through JavaScript. A page Claude-User cannot read in one request is a page Claude cannot quote.

2. Be visible in Brave Search

Search your brand name and your key questions on Brave Search and note which of your pages appear. Brave runs its own index and offers no public submission console comparable to Google Search Console, so discovery comes from links and visits: a clean internal structure, an up-to-date sitemap and links from other sites remain the levers. Pages that rank well on Google usually do well on Brave too, because the same quality signals apply.

3. Write answer-first passages

Claude reads passages, not whole pages. Open each section with a direct answer of two or three sentences, use headings phrased like the questions your buyers ask, and keep one idea per section. Lists, tables and short definitions survive being cut out of the page; vague introductions do not. The research paper GEO: Generative Engine Optimization (Aggarwal et al., 2023) found that adding sources, quotations and statistics raised a page's visibility in generative answers by up to 40%.

4. Give Claude facts it can check

Claude is trained to be cautious with claims it cannot support, so dated figures, named sources, prices, specifications and author names make a page easier to cite. Write "checked October 2026" next to a price, link to the primary source of a statistic, and avoid superlatives you cannot prove.

5. Build consensus beyond your site

Claude trusts what several independent sources agree on. Reviews on the platforms of your sector, mentions in comparisons, directory listings and articles in specialized media all reinforce the same description of your brand, which is what entity consensus means. Natural backlinks from relevant sites help on both Brave and Google.

6. Keep key pages fresh

Web search exists to answer what training data cannot: recent prices, new products, current rankings. Update your comparison, pricing and statistics pages with a visible date, and publish regularly on the questions your market asks. Our article generator writes and publishes sourced, structured articles on those questions on autopilot.

How do you measure your visibility in Claude?

Claude does not offer a console that shows impressions or citations, so measurement combines three signals:

• Prompt tests: list 20 to 50 questions your buyers ask, run them in Claude with web search on, and note whether your brand is mentioned, whether your pages are cited and who is cited instead.
• Server logs: visits from Claude-User show pages fetched live for a real question, the closest signal to a citation, as explained in our entry on AI crawler logs.
• Referral traffic: visits from claude.ai in your analytics measure the clicks those citations bring.

Answers vary from one run to the next, so a single test proves little. Tracking the same questions every week is what turns impressions into a trend. Sorank's AI mention tracking runs your prompts on Claude, ChatGPT and Gemini on every plan, and on Perplexity and Grok from the Pro plan, with the sources each answer cites. The dedicated Claude rank tracker page explains the Claude view.

What does ranking in Claude change for SEO and GEO?

Claude rewards the same foundations as Google: crawlable pages, clear answers, verifiable facts and a reputation confirmed by other sites. What changes is the scarcity. With about three to four sources per answer, being the second-best page on a question often means not being cited at all. The sites that win in Claude are the ones that answer precisely, prove what they say and are confirmed elsewhere. To see where you stand today, compare your plans on the Sorank pricing page and start tracking the questions that matter to your business.