How to get indexed in Brave Search (and why Claude cites it)

What “ranking in Brave Search” depends on

Brave Search runs its own index, and Brave says its crawler is partly powered by pages that opted-in Brave browser users visit through the Web Discovery Project, or WDP (Brave Search Crawler). WDP is off by default, and Brave states that the data it receives is unlinkable to users (Brave announcement). So a page enters the index through its own crawler and through user signals, not through a console where you register a site.

Why care: Claude’s web search is widely reported to use Brave results, and an independent study of about 35,000 Claude citations found 79.2% of cited URLs inside Brave’s top 10 for the matching query, against 34% for Google’s top 10 (MERJ analysis). That is a third-party measurement, not an Anthropic statement, so treat it as evidence of correlation, not a guarantee that a Brave ranking produces a citation.

What the “Submit URL” form actually does

Brave’s help page offers a link to submit a page for re-fetching, described for the case where you changed a noindex directive, and a separate one for URLs that were delisted. It does not describe a way to submit a new site for discovery. Use it after a fix, not as a growth tactic.

Two ways a page gets discovered

The independent source-code analysis of the open-source WDP client describes two channels. Brave’s servers are not open, so the server side is inferred, and the numbers below are the study’s findings, not Brave’s published promises.

The practical reading is that a page already ranking in another engine has an easier way in than a brand-new page with no visitors. If you have no rankings yet, do the ordinary work first: crawlable pages, internal links and real promotion.

What the verification fetch can see

Before a page is reported, the WDP client re-fetches it anonymously to check that it looks the same for everyone. The analysis found these rules in the client code.

1. No JavaScript. The fetch reads raw HTML, so content that exists only after client-side rendering is not there.
2. No redirects. Any redirect means the page is treated as private and not reported.
3. 10 seconds and 2 MB. The fetch aborts after 10 seconds, and the size check uses the content-length header with a 2 MB limit.
4. HTML signals only. noindex counts when it appears in the HTML <head>; no X-Robots-Tag support was found in the code. The canonical comes from a <link rel="canonical"> element, not from a Link: header.
5. Query strings. Without a canonical, a second query parameter or a query string longer than 22 characters marks the page private.
6. Path words. Paths containing words such as login, token, checkout, account, admin or wp-admin are rejected.

These rules describe one component, the WDP verification fetch. They do not describe everything Brave’s own crawler does, and they can change.

Do not block the crawler by accident

Brave says its crawler does not advertise a distinct user agent, to avoid discrimination against sites that allow only Google, and that a page not crawlable by Googlebot will not be crawled by Brave either. A blanket block on unknown bots in robots.txt or in a firewall or CDN rule can therefore exclude it silently. Note also that robots.txt is not used to keep a page out of the index; for that Brave points to the noindex directive, as Google does (block indexing with noindex, robots.txt introduction).

A 20-minute audit

1. Read robots.txt. Look for a catch-all Disallow or rules that allow only named crawlers; then check firewall and CDN bot rules.
2. View source, not the rendered page. Confirm the main text is in the raw HTML.
3. Trace one URL. It should return 200 with no redirect, in well under 10 seconds and under 2 MB.
4. Check the head. A noindex or canonical sent only as an HTTP header is not seen by the verification fetch; put it in the HTML as well.
5. Clean the canonical. Use a canonical URL without needless parameters and without risky path words (consolidate duplicate URLs).
6. Look for existing rankings. Pages that already appear on Google, Bing or DuckDuckGo have the easier channel.

Should you use the Brave browser yourself?

If you opt in, your visits to your own public pages could count as one of the reports in the page channel, and searching a query where the page already ranks elsewhere could trigger the query channel. It is cheap, but it is one signal among about 20, so it is not a plan. To opt in, open Settings, choose Search engine and switch on Web Discovery Project (Brave help).

What this playbook cannot promise

• No guaranteed discovery. Brave publishes no submission path that guarantees indexing.
• No guaranteed citation. A study measured overlap between Claude citations and Brave results; it did not show that ranking in Brave causes a citation.
• Inferred server behaviour. The analysis verified client code, and the server side is inferred.
• Moving rules. Details such as limits and thresholds can change; re-check before relying on one.

Brave Search indexing, answered

Can I submit my site to Brave Search?

Brave’s help page only offers a form to ask for a re-fetch of a page, for example after you change noindex. It does not describe a way to submit a new site.

Does Claude use Brave Search?

It is widely reported, and a third-party study found a large overlap between Claude’s cited URLs and Brave’s top results. We could not verify an Anthropic statement, so we say “reported”.

Does Brave have a search console?

We found no Brave tool that works like Search Console. Check your own server logs and rankings instead.

Does robots.txt keep a page out of Brave?

No. Brave says robots.txt is not used to prevent indexing; it points to noindex.

What to read next

The crawlability problems that hide a page from Brave usually affect other engines too. Go back to the technical indexing audit or open the full learning track.