How Does Claude Decide Which Sources to Cite?
·
Quint-IA Vantage Team
Claude, the AI assistant made by Anthropic, searches the web when a question needs current information, and cites the pages it uses. Unlike Gemini, which leans on Reddit, YouTube, and Wikipedia, Claude cites company websites most of all.
In a study of 379,321 Claude citations on SaaS and technology queries in June 2026, Otterly.AI found that 64% pointed to brand websites [4]. Community forums made up 1.3%, and Reddit was not cited once.
That makes Claude the answer engine where your own site carries the most weight. This piece explains when Claude searches, which Anthropic crawlers reach your pages, how citations work, and what to change on your site. It draws on Anthropic's own documentation and two independent citation studies.
When does Claude search the web?
Claude searches when a question depends on information that is current or outside what the model learned in training. Anthropic's developer documentation says Claude decides when to search, and answers directly from its own knowledge for stable facts, math, or creative writing [1].
In the Claude apps, web search is a setting the user turns on. Once it is on, Claude runs a search when a question benefits from current information, and users can also ask it to search directly [2].
Claude can search more than once in a single answer. The documentation describes progressive searches, where Claude uses earlier results to decide what to search next [1]. Claude's Research feature, available on paid plans, runs many searches that build on each other and explore different angles of a question [3].
Anthropic's documentation does not name the search index behind Claude's web search. Some third-party analyses report that Claude's citations closely match Brave Search results [5], but Anthropic does not confirm this on its current help or developer pages.
Which Anthropic crawlers visit your site?
Anthropic runs three crawlers (automated bots that read web pages), and each has a different job. All three follow robots.txt rules, the file that tells crawlers which pages they may visit [6].
ClaudeBot collects web content that may be used to train Anthropic's models. Blocking it excludes your future content from training data.
Claude-SearchBot indexes content to improve Claude's search results. Blocking it may make your pages less visible in Claude's search results.
Claude-User visits pages when a person asks Claude a question. Blocking it stops Claude from retrieving your pages to answer user questions.
Source: Anthropic Help Center [6].
The split matters for your robots.txt. Blocking ClaudeBot affects model training only. Blocking Claude-SearchBot or Claude-User reduces your chances of being found and cited when someone asks Claude a question, according to Anthropic.
This differs from Perplexity, whose user-triggered agent generally ignores robots.txt. Anthropic states that all three of its crawlers respect it.
How does Claude cite its sources?
When Claude answers with web search, it attaches citations to the claims that came from the web. Anthropic's developer documentation says citations are always enabled for web search, and each one carries the source URL, the page title, and up to 150 characters of the quoted text [1].
In the Claude apps, users see these as direct citations and source links they can click to check the original [2]. Developers who show Claude's web search answers to their own users must keep the citations to the original sources [1].
The 150-character limit has a practical consequence. Claude quotes a short passage, so the sentence that answers the question needs to stand on its own. An answer spread across several paragraphs is harder to capture in a short excerpt.
Which sources does Claude cite?
Claude cites company websites far more than any other type of source. Otterly.AI analyzed 379,321 Claude citations across 16,406 domains on SaaS and technology queries in June 2026. Brand websites accounted for 64.0% of citations, news and media sites for 14.9%, and blogs for 5.4% [4].
Source: Otterly.AI, June 2026, SaaS and technology queries. Remaining 3.7% classified as other [4].
Citations were also spread thin. The 10 most-cited domains accounted for just 9.5% of all citations, and a third of cited domains appeared only once [4]. There is no small group of sites that Claude defaults to.
Reddit received no citations in the dataset, and video made up 0.3% [4]. Gemini's three most-cited domains, by comparison, are Reddit, YouTube, and Wikipedia [7].
A second, smaller analysis points the same way. Oltre.ai reviewed 2,170 URLs cited by Claude and found that 56% sat under /blog/ paths, 47% were list or comparison pages, and only 3% were homepages [5]. The cited pages are specific articles that answer a question, and homepages are rare.
Claude also draws on different sources than ChatGPT. Only 13% of the domains Claude cited were also cited by ChatGPT, and about 4.2% overlapped at the URL level [4].
One limit to keep in mind: the Otterly data covers SaaS and technology queries only. Citation patterns for other industries may differ.
What should you do to get cited by Claude?
Focus on your own website first, since that is where most of Claude's citations go. Five steps follow from how Claude searches and cites.
1. Check your robots.txt for all three Anthropic crawlers. Make sure Claude-SearchBot and Claude-User are allowed, and check your CDN or firewall rules too [6]. You can block ClaudeBot if you do not want your content used for training, without blocking the two crawlers that affect search.
2. Publish focused articles for each buyer question. Claude's citations go to specific blog posts, guides, and comparisons far more than homepages [5]. Otterly recommends building a library of focused pages rather than a few broad ones [4].
3. Write sentences Claude can quote on their own. Claude's citations carry a short excerpt of up to 150 characters [1]. Put a clear definition or direct answer at the top of each section, and back it with a specific number or source.
4. Earn coverage in relevant news and industry outlets. News and media sites are Claude's second-largest source of citations, at 14.9% [4]. Coverage in outlets your buyers read gives Claude a second source for the same facts.
5. Track Claude separately from ChatGPT. With only 13% of cited domains in common between the two [4], strong visibility in one tells you little about the other.
What this means for your AI search plan
Of the answer engines covered in this series, Claude gives the most weight to what you publish on your own site. It searches when a question needs current facts, respects robots.txt, quotes short passages, and spreads its citations across thousands of company domains. A clear, well-structured library of articles on your site is the most direct route into its answers.
Frequently asked questions
Does Claude search the web? Yes, when web search is turned on. Claude decides whether a question needs current information and searches when it does, sometimes several times in one answer. For stable facts, math, or writing tasks, it answers from what the model already knows. Answers that use web search include citations linking to the original pages.
What is the difference between ClaudeBot, Claude-SearchBot, and Claude-User? ClaudeBot collects web content that may be used to train Anthropic's models. Claude-SearchBot indexes pages to improve Claude's search results. Claude-User visits pages when a person asks Claude a question. Anthropic says all three respect robots.txt, and blocking the last two can reduce your visibility in Claude's answers.
Can I block Claude from training on my site but still appear in its answers? Yes. Anthropic's crawlers are separate, so you can block ClaudeBot in robots.txt to opt out of training while allowing Claude-SearchBot and Claude-User. Blocking ClaudeBot signals that your future content should be excluded from training data, according to Anthropic.
What kinds of websites does Claude cite most? Company websites. In Otterly.AI's study of 379,321 Claude citations on SaaS and technology queries in June 2026, brand websites made up 64% of citations and news sites 14.9%. Community forums were 1.3%, and Reddit received no citations at all.
How is getting cited by Claude different from ChatGPT or Gemini? Each engine draws on different sources. Only 13% of the domains Claude cited were also cited by ChatGPT in Otterly's data. Gemini leans on Reddit, YouTube, and Wikipedia, while Claude cites company websites most. A page that wins citations in one engine may not appear in another.
Which search engine does Claude use? Anthropic's help and developer documentation does not name the search provider behind Claude's web search. Some third-party analyses report a close match between Claude's citations and Brave Search results, but Anthropic has not confirmed this on its current documentation pages.
See how your brand scores across 4 answer engines.
References
Anthropic. "Web search tool." Claude Developer Platform documentation. platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool
Anthropic. "Enabling and using web search." Claude Help Center. support.claude.com/en/articles/10684626
Anthropic. "Use Research on Claude." Claude Help Center. support.claude.com/en/articles/11088861
Tousseyn, R. "Claude AI Citations Study." Otterly.AI, published June 30, 2026 (data from June 2026). otterly.ai/blog/claude-ai-citation-study
Pizzola, L. "How Claude Picks Sources: A Technical Breakdown." Oltre.ai, published May 18, 2026. oltre.ai
Anthropic. "Does Anthropic crawl data from the web, and how can site owners block the crawler?" Claude Help Center. support.claude.com/en/articles/8896518
Lafferty, N. "How to Rank and Get Cited in Google's Gemini." Profound, published September 4, 2026. tryprofound.com/articles/rank-in-google-gemini