Meet an Expert

Why Claude Runs Web Search Through Brave, Not Google

Claude's web mode runs on Brave Search, not the open web, which changes which brands get cited and how to optimize for it.

Lior Eldan Lior Eldan
Oct 23, 2026 6 min read Engines

Ask ChatGPT and Claude the same question about enterprise payroll software. Compare the answers. Often they name entirely different companies. ChatGPT draws from a broad crawl of the open web. Claude’s web mode works differently: it routes the query through Brave Search, a separate index with its own crawl priorities and editorial leanings. The result is a citation mix that doesn’t match ChatGPT’s or Gemini’s, and a brand built for one is often invisible in the other.

Key Takeaways

  • Claude’s web mode retrieves through Brave Search, not a general web crawl, so its citation mix reflects what Brave indexes and ranks, which is a different set of sources than ChatGPT or Gemini favor.
  • Ranking well on Google says nothing about whether a page exists in Brave’s independent index, so pages built only for Google Search Console can be structurally absent from Claude’s retrieval pool.
  • Share of Citation on Claude can diverge sharply from Share of Citation on ChatGPT or Gemini, because the retrieval layer underneath the model, not the model itself, decides what gets surfaced first.

Why Claude Sees a Different Web

Every engine in this category has a different relationship with the web. ChatGPT browses across the open web. Gemini cross-references Search, YouTube and Scholar, pulling in video transcripts and academic papers that the others rarely touch. Claude leans on high-authority publications and documentation when it reasons, and once its web mode goes live, Brave Search is the engine doing the actual fetching.

Brave runs its own index, built by its own crawler. It doesn’t license results from Google or Bing the way plenty of smaller search products do. That independence matters more than it might sound. A page sitting on page one of Google for a given query isn’t automatically anywhere in Brave’s index. And if it isn’t indexed, Claude’s web mode has nothing to cite.

What Gets Indexed, and What Doesn’t

Brave Search leans toward independent publishers, structured documentation and sites with clear ownership. Forum threads and aggregator content, the stuff that often dominates Google’s results page, get less weight here. For a model that already favors high-authority publications and documentation in its training data, retrieval through Brave doesn’t broaden that preference. It reinforces it. The practical effect: a narrower, more editorial slice of the web feeds every Claude web-mode answer.

Citations compound once this pattern starts. A source that Brave surfaces and Claude cites for one query becomes more likely to get cited again on an adjacent query. Trust here behaves like a graph of agreement across independent sources, not a single score handed down from one domain. The same compounding logic shows up in our own tracking of AI citations across engines.

309
unique pages of Moburst’s own site cited across AI engines, part of a dataset showing 42,435 total AI citations and the number 1 position in its category.See the case study

Same Question, Different Engine, Different Winner

A brand with strong tutorial coverage on YouTube can show up reliably in Gemini’s answers, since Gemini cross-references YouTube directly. Ask Claude the same thing and that brand might vanish entirely, because Brave doesn’t weight video content the way Gemini’s own index does. A brand with heavy discussion volume on social platforms might surface in Grok instead, which reads the real-time social web, while staying invisible to Claude’s documentation-leaning retrieval.

Perplexity sits somewhere between these, citing its sources inline and rewarding content that answers a question in a self-contained way regardless of format. Claude’s bias runs toward the kind of page a careful editor would link to: a documentation page, a published report, a site with a clear author and a clear owner.

Claude does not read the whole internet, it reads what Brave Search decided was worth indexing, and that is a narrower, more editorial slice of the web.

What This Means for Pages Built for Claude Web Mode

Share of Voice measures how often a brand gets mentioned across the web. Share of Citation measures something different: how often a model actually pulls that brand in and names it as the source. Claude’s narrower retrieval path widens the gap between the two. Fewer opportunities to get picked up means each one carries more weight.

A page built for Claude needs to exist in Brave’s index first. That means checking crawlability for Brave’s crawler specifically, not assuming that Google access already covers it. After that, the content itself needs the shape Brave and Claude both reward: clear structure, a direct answer near the top, visible authorship, and corroboration from other independent sources rather than a single self-published claim repeated across owned channels.

What to Do Next

Search your own priority queries directly in Brave Search, not just Google, and note which of your pages even appear. If a page is missing from Brave’s results, it has little chance of reaching a Claude web-mode answer, no matter how well it performs elsewhere.

Check whether Brave’s crawler has unrestricted access to the pages you care about. A robots.txt rule written years ago for a different purpose can quietly block it without anyone noticing. Build or tighten documentation-style pages for your core topics: a clear question, a direct answer, a visible date, a named author. Pursue coverage on independent, high-authority publications rather than volume of generic mentions, since trust here is agreement across sources, not output from one domain.

And track citation mix per engine instead of one blended “AI visibility” number. A Model Coverage Map that separates Claude from ChatGPT and Gemini will show you exactly where the gap sits, rather than hiding it inside an average.

FAQs

Why Does Claude Show Different Sources Than ChatGPT for the Same Question?

Claude’s web mode retrieves through Brave Search’s independent index, while ChatGPT draws from a broader open-web crawl. Different retrieval infrastructure produces a different set of candidate sources before either model writes an answer.

Does Ranking Well on Google Guarantee a Page Appears in Claude’s Answers?

No. Brave Search runs its own index and crawler, built independently of Google and Bing, so a page’s Google ranking has no direct bearing on whether Brave has crawled or ranked it at all.

What Is the Difference Between Share of Citation and Share of Voice Here?

Share of Voice measures how often a brand is mentioned across the web overall. Share of Citation measures how often a model actually pulls a brand in and names it as a source. A brand can score high on one and barely register on the other, and Claude’s narrower retrieval through Brave makes that gap more visible than it is on ChatGPT or Gemini.

What Should a Brand Check First for Visibility in Claude Specifically?

Search the target queries directly in Brave Search to see whether the page is indexed and ranked there at all. If it does not appear in Brave, it has little chance of being pulled into a Claude answer, regardless of how it performs on Google.

Lior Eldan

About the author

Lior Eldan

Lior Eldan is the Co-Founder of Moburst and serves as its COO. He works at the intersection of marketing, AI and growth, helping brands' teams adapt to AI-driven discovery and decision-making through data-informed strategy and systems thinking.

Meet the whole team

Ready to make your brand the answer?

Our growth strategists take brands from overlooked to recommended. In one call you will see how the AI engines read your brand today, and leave with your own roadmap to dominate the new era of search.