TL;DR — On June 12, 2026, Search Engine Land published analysis showing that Claude, Anthropic's assistant, builds its web answers almost entirely from Brave Search's top 10 results. Its citations align with those results around 87% of the time. The direct consequence: to get cited by Claude, you need to rank in Brave's organic top 10. It is, to date, the most predictable and optimizable AI answer engine.
On June 12, 2026, journalist Danny Goodwin published an analysis on Search Engine Land revealing how Claude selects the sources it cites. The data comes from Jonathan Clark, managing partner at Moving Traffic Media, who shared the results of a "Zero Click" session run with the Profound tool on LinkedIn. The finding is clear: when Claude searches the web, it leans on Brave Search, and it almost never re-ranks those results.
In practice, Claude doesn't pass the user's raw question to Brave. It rewrites it into a search-optimized format, retrieves the top ten organic results (no sponsored results), then writes its answer from that list. Its final citations match Brave's top 10 around 87% of the time.
The numbers that matter
Claude doesn't search for everything. According to the data shared by Jonathan Clark, it goes to the web mostly when the prompt signals a need for freshness, ranking or comparison:
- Freshness prompts ("latest", "2026"): search triggered 81% of the time
- Ranking prompts ("best X", "top tools"): 67%
- Location prompts: 55%
- Comparison prompts ("X vs Y"): 51%
Another important signal: Claude's query fan-outs (how it breaks a question into several searches) are nearly deterministic 65% of the time. In other words, its behavior is largely reproducible, and therefore testable.
Why Claude is different from other AI
The most telling figure is the 8% overlap with ChatGPT. The two assistants barely cite the same pages. Where ChatGPT searches in roughly 90% of prompts and draws from its own index, Claude delegates its search to Brave and stays faithful to its ranking. Its proximity to Google (64%) is still stronger than to ChatGPT, but it's Brave that decides what ends up in the answer.
According to the analysis, Brave reworked its algorithm to reward not just relevance but also clarity, neutrality and citation quality. Clear, factual, non-overselling content therefore has a mechanical advantage in reaching Brave's top 10, and thus Claude's answers. It's the same underlying shift we described in the fragmentation of market share across ChatGPT, Gemini and Claude: each AI engine builds its own citation logic, and you have to treat them separately.
The Cicéro angle — Most companies optimize for Google and, recently, for ChatGPT. Almost no one tracks their Brave Search ranking. That's exactly where there's a largely uncontested window of AI visibility.
What it changes for your visibility
If some of your prospects use Claude to compare providers or find "the best X in 2026", your presence in its answers now depends on a measurable lever: your position on Brave Search. Three concrete actions follow.
- Track your Brave Search ranking. Brave is a fraction of the search market, but it commands visibility inside Claude today. Add a few key queries to your rank tracking.
- Add current-year signals to your titles. Freshness prompts trigger a search 81% of the time. A title with "2026" has a clear edge on news- and ranking-driven prompts.
- Produce comparison and news formats. These are exactly the intents that push Claude to search. "X vs Y" pages, "top tools" rankings, dated analysis: those are your best entry points.
This logic extends what we wrote about zero-click search and Google's official guide to optimizing for AI: visibility no longer plays out on Google alone, but across every answer engine, each with its own rules.
What this article does not cover
These figures come from a sample of prompts analyzed via Profound and shared by Jonathan Clark; they show a strong trend, not a fixed guarantee. Claude's behavior can shift with every Anthropic or Brave algorithm update. This article also doesn't lay out a step-by-step Brave rank-tracking method, nor the non-web case (Claude also answers without searching in 63% of prompts, from its training). One to watch in the coming weeks.
Our take
While everyone fights to decode ChatGPT's black box, Claude offers a rare shortcut: an answer engine whose source is known and publicly ranked. Working your Brave Search ranking is, today, one of the most direct GEO levers that exists. The window is open, and it won't stay open forever.
Frequently asked questions
How does Claude choose its web sources?
How often does Claude actually search the web?
How do you get cited by Claude?
Sources
- → Search Engine Land — "Claude visibility may depend heavily on Brave Search rankings", Danny Goodwin, June 12, 2026.
- → Original data: Jonathan Clark (Moving Traffic Media), "Zero Click" session via Profound, shared on LinkedIn (June 2026).
- → Brave Search — engine used by Claude's search API.
Growth and SEO content strategist, I founded Cicéro to help businesses build lasting organic visibility — on Google and in AI-generated answers alike. Every piece of content we produce is designed to convert, not just to exist.
LinkedIn