Claude Referral Traffic Is Rising: How to Get Cited by Claude

To get cited by Claude, make sure Claude-SearchBot and Claude-User can reach your pages, publish short answer-first passages with specific facts, keep dates and details current, and track Claude visits in GA4. Claude searches the web when a question needs fresh information and links the pages it uses, so the basics of crawlability and clarity matter most.

Until recently, few small businesses paid attention to Claude as a traffic source. That is changing quickly. This guide shows the data behind the rise, how Claude's web search works according to Anthropic's own documentation, and a practical checklist you can work through in an afternoon.

Claude's share of AI referrals has jumped

Trakkr's AI search traffic index, built from GA4 data across a panel of about 6,600 websites, shows how quickly the picture changed in 2026:

Share of identifiable AI referral page views. Source: Trakkr, retrieved November 2026.

In the 30 days to November 4, 2026, Trakkr puts Claude at 17.4% of AI referral page views, up from 12.3% a month earlier, with Claude page views up 51.7% over that period.

Keep the numbers in context:

• These are shares of AI referrals, not of all traffic. Trakkr says its index is not global market share and does not measure total AI search volume. Clicks that lose their referrer show up as Direct and are excluded.
• Other panels show smaller absolute numbers. SE Ranking's study of 101,574 websites found Claude's referral traffic grew 386% between January and April 2026, but from 0.0029% to 0.0141% of total website traffic. Growth is fast; the base is still small for most sites.
• Methodology varies, so compare trends within one source rather than mixing figures from different studies.

The practical takeaway: Claude is now worth a line in your AI search reporting and a check of your robots.txt, especially if you sell to businesses or technical buyers.

How Claude finds and cites web pages

Anthropic launched web search in Claude in March 2025 and, in a May 27, 2025 update, said it "is now available globally on all Claude plans." When Claude uses web information, Anthropic says it "provides direct citations so you can easily fact check sources."

Anthropic's developer documentation for the web search tool gives more detail on how the same capability behaves for developers building on Claude. It is the clearest public description of what Claude looks for:

• When Claude searches. It searches when a request depends on information that is current, changing or outside its training data, such as recent news, current prices, rates or statistics, and details about specific organizations, people or products that might have changed. It answers directly, without searching, for stable knowledge such as established facts or coding concepts.
• What it gets back. Each search result includes the page URL, title and page_age, described as when the site was last updated.
• How citations work. Citations are always enabled for web search. Each citation includes the URL, title and cited_text, which is up to 150 characters of the cited content.
• Several searches per answer. The documentation says single-fact queries typically use one to three searches, while comparative research can use ten or more.

Two things follow for site owners. Claude is most likely to look you up for questions about things that change, such as your prices, opening hours, services, availability or recent news. And the part of your page Claude quotes is short, so a clear sentence beats a long, winding paragraph.

The three Claude crawlers

Anthropic's crawler help article describes three bots, and they do different jobs:

Anthropic says its bots honor robots.txt and that it supports the non-standard Crawl-delay directive. It advises against IP blocking, saying it "may not work correctly or persistently guarantee an opt-out."

You can block ClaudeBot and still allow Claude-SearchBot and Claude-User. That combination opts out of training while keeping you eligible for citations. Our guide on whether to block or allow AI crawlers covers the same split for OpenAI, Google, Perplexity and others.

Checklist: how to get cited by Claude

1. Check that Claude's search bots can reach you

Open your robots.txt (yoursite.com/robots.txt) and look for rules naming Claude-SearchBot, Claude-User, anthropic-ai or a blanket User-agent: * with Disallow: /. Also ask your host or CDN whether a bot-blocking or firewall feature is switched on, because a server-level block stops a crawler regardless of what robots.txt says.

The free AI crawler checker shows which AI bots your robots.txt allows, and the robots.txt tester lets you test a specific URL against a specific user agent. Both need a free account.

A minimal setup that allows Claude search while opting out of training:

User-agent: ClaudeBot
Disallow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: Claude-User
Allow: /

2. Put the answer where a quote can find it

Since citations carry about 150 characters of text, write at least one sentence per page that answers the main question on its own. For example, a plumber's pricing page might open with one sentence stating the call-out fee, the city it applies to and what it covers. That sentence makes sense without the rest of the page.

Practical habits:

• Start each page or section with a direct answer, then explain.
• Use question-style headings that match how people ask, then answer in the first sentence below.
• Put key facts in text, not only in images, PDFs or tabs that load later.

The heading checker helps you spot pages where the structure hides the main answer.

3. Keep changing facts current and dated

Claude searches most for current, changing information, and each search result carries a page_age. Make sure the pages that hold changeable facts are accurate:

1. Prices and fees
2. Opening hours, service areas and contact details
3. Product availability and specifications
4. Policies such as returns, cancellation or booking terms
5. Team, credentials and licenses

Show a visible "last updated" date on those pages and change it only when the content really changes.

4. Make the page easy to fetch

When Claude-User fetches a page for a user, a slow or script-heavy page is a risk. Keep the main content in the HTML, avoid putting key text behind logins or cookie walls, and make sure the page returns a normal 200 status to bots. If you rely on client-side rendering, test that the content appears in the page source.

5. Be specific about who you are

Questions about "specific organizations, people or products" are exactly the kind Claude searches for. Have a clear About page, consistent business name, address and phone across your site and profiles, and plain statements of what you do and where. Our guide to getting your business mentioned by ChatGPT covers the same entity basics, and they apply to Claude too.

6. Track Claude visits separately

In GA4, open Traffic acquisition and filter session source for "claude". Better still, create a custom AI channel group that matches sources such as claude, chatgpt and perplexity and sits above Referral, as described in our article on the October 2026 ChatGPT referral traffic drop. Remember that some AI clicks arrive without a referrer and count as Direct, so referral numbers are a floor.

7. Test a fixed set of prompts monthly

Write down 10 to 20 questions a customer might ask Claude where your business should appear, such as "best accountant for Etsy sellers in Ohio" or "how much does a roof inspection cost in Phoenix". Ask them once a month with web search on, and record whether you are cited and which competitors are. It is manual, but free and reliable. The AI visibility checker gives a quick technical baseline alongside it.

What not to waste time on

• Hidden text or keyword lists for AI. Claude quotes what a reader would see; content written only for bots is unlikely to be the 150 characters it picks.
• Assuming llms.txt is required. We found no Anthropic documentation saying Claude's search reads llms.txt. It may help some tools; see what llms.txt is before investing time.
• Blocking by IP address. Anthropic recommends robots.txt instead.

Frequently asked questions

How does Claude decide which websites to cite?

Anthropic's documentation says Claude searches when a question depends on current or changing information, reviews the results and cites the sources it draws on. Each citation includes the page URL, title and up to 150 characters of quoted text. Pages that are reachable, clearly written and up to date give it the most to work with.

Does Claude use web search for every question?

No. According to Anthropic's documentation, Claude answers directly for stable knowledge such as established facts, math or coding concepts, and searches for recent events, current prices or statistics, and details about specific organizations or products that may have changed.

Should I block ClaudeBot?

That depends on whether you want your content used for model training. Blocking ClaudeBot excludes your future content from Anthropic's training data. It does not block Claude-SearchBot or Claude-User, which are the bots tied to Claude's search results and user-requested fetches.

What is Claude-SearchBot?

Claude-SearchBot is Anthropic's crawler that indexes web content to improve Claude's search results. Anthropic says disabling it may reduce your site's visibility and accuracy in user search results. It is controlled separately from ClaudeBot in robots.txt.

How do I see Claude traffic in Google Analytics?

In GA4, go to the Traffic acquisition report and filter session source for "claude". For ongoing reporting, create a custom channel group with an AI assistants channel that matches Claude, ChatGPT, Perplexity and Gemini sources. Some Claude clicks may arrive without a referrer and appear as Direct.

Is Claude traffic bigger than Perplexity traffic now?

In Trakkr's GA4 panel, Claude accounted for 17.4% of AI referral page views in the 30 days to November 4, 2026, compared with 0.5% for Perplexity. Other studies measure differently and show different shares, so check your own analytics to see what applies to your site.

Sources

• Trakkr: AI Search Traffic live index (retrieved November 5, 2026)
• SE Ranking: What Claude's Automation Boom Means for AI Visibility (June 3, 2026)
• Claude blog: Claude can now search the web (March 20, 2025, updated May 27, 2025)
• Claude Developer Platform: Web search tool documentation
• Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?