Yes, Claude recommends law firms, but only when a user’s question calls for fresh information and only from the handful of sources it pulls during a live web search. Since March 20, 2025, Claude has searched the web through Brave Search, retrieves roughly 10 results per query, reads each one, and cites the few it trusts with inline links. If your firm is not in Brave’s index, or your pages do not answer the exact question a prospect asks, Claude never sees you. This is the AI engine most law firms ignore, and that is the opening.
That is the short version. Here is how Claude actually picks legal sources, why the Brave backend changes your playbook, and the three moves that get a firm cited.
How Claude decides whether to recommend a law firm at all
Claude does not answer every question with a web search. For a definitional query like “what is a contingency fee,” it answers from training data and never searches. For a query with local or commercial intent, like “best estate planning attorney in Charleston” or “who are the top personal injury firms near me,” Claude triggers its search tool, pulls live results, and builds the answer on top of what it finds.
This split matters. The queries that drive cases are almost always the ones that trigger a search, because they involve a specific place, a recent fact, or a comparison Claude cannot answer from memory. So the firms that win on Claude are not the ones with the most training data exposure. They are the ones that show up in the live search and survive Claude’s source assessment.
When Claude does search, it shows the user the search terms it used, lists the underlying results, and then writes its answer with citations and clickable source links inline. The user sees your firm name next to the answer even if they do not click. That is brand impression in the highest-trust surface a buyer touches during research.
Claude runs on Brave Search, and that rewrites your indexing strategy
Here is the fact most AEO advice for law firms misses. Claude does not use Google’s index. Anthropic added Brave Search to its subprocessor list, and independent testing found an 86.7% overlap between Claude’s cited results and Brave’s top non-sponsored results. Claude reads the web through Brave’s eyes.
That has a hard consequence. A page becomes eligible for Brave’s index only after at least 20 unique Brave browser users with data sharing enabled have visited it. Brave builds its index from real browsing activity, not from a crawler that visits every URL on the open web. So a brand new practice area page can rank beautifully on Google and stay invisible to Claude for weeks, because Brave has not picked it up yet.
The practical read for a law firm: ranking on Google is necessary but not sufficient. You also need Brave to know your pages exist. The fastest way to force that is to make sure your highest-value pages get real traffic, get submitted through Brave’s tools, and earn links from sites Brave already crawls. Most firms have never checked whether their pages are in Brave’s index at all. Searching your firm name and top practice areas directly in Brave Search takes five minutes and tells you exactly where you stand.
What Claude rewards once your pages are in the index
Being in the index gets you into the candidate pool. Getting cited is a second test. Claude reads each of the roughly 10 results it pulls and scores them for quality, relevance, and trust before it writes a single word. A few patterns separate cited pages from skipped ones.
Claude favors structure. Testing shows Claude is about 30% more likely to cite a page that uses clear bullet points and labeled sections than a page that buries the same facts in dense paragraphs. It prefers long form, well organized guides over thin pages.
Claude rewards traceable evidence. Anthropic trained the model to prize first-party data, clear declarative claims, and statements backed by primary sources. A practice area page that says “we have handled more than 400 wrongful death cases since 2009” beats a page that says “we have extensive experience,” because the first is a checkable claim and the second is filler.
Claude filters hard on trust. Claude is built on Constitutional AI with strong safety guardrails, and content that reads as manipulative, exaggerated, or factually shaky gets discounted or dropped. For law firms, that means no overclaiming, no fake urgency, and accurate, consistent information across every page. Ethical, well-sourced writing is not a nice-to-have here. It is a precondition for visibility.
And Claude weights entity consistency. When your firm name, address, phone number, and attorney credentials match across your site, Google Business Profile, Avvo, Martindale-Hubbell, Justia, and your state bar profile, Claude reads that consistency as a trust signal. When the details conflict, it reads risk and pulls a competitor instead.
Why Claude’s research mode raises the bar for law firms
Claude’s research and agentic modes run multi-step analysis across many sources before answering. Instead of one search, Claude fires several sub-queries, reads dozens of pages, and synthesizes a single answer. Anthropic’s February 9, 2026 search update went further and lets Claude write and run Python to process raw page HTML before it reaches the model, which means it can parse tables, structured data, and messy pages more reliably than before.
For a law firm, this favors brands that appear across multiple high-quality domains rather than firms with one strong page. If Claude reads your firm on your own site, then sees you corroborated on Justia, Avvo, a state bar profile, and a piece of editorial press, the repeated independent confirmation is what convinces it to put your name in the answer. One great page is a candidate. The same firm verified across five trusted domains is the citation.
Claude is now inside law firms, which makes consumer visibility more urgent
In May 2026, Anthropic launched Claude for Legal with plugins and connectors for legal research, document review, and deposition prep, plus a Westlaw integration. Lawyers are already using Claude inside their own practices every day. That same comfort carries over to how their clients and referral sources behave. When a general counsel or a prospective client reaches for an AI assistant to vet outside counsel, Claude is increasingly the tool in hand. Being the firm Claude names in that moment is worth more than any single Google ranking.
The three moves that get your firm cited by Claude
First, confirm Brave indexing and force it where pages are missing. Search your firm name and your top three practice area queries directly in Brave. Any page that does not appear is invisible to Claude regardless of its Google rank. Drive traffic and links to those pages until Brave picks them up.
Second, restructure your top pages into question-and-answer format. Make the H1 the exact question a buyer asks. Answer it completely in the first 40 to 80 words. Make every H2 a related sub-question with its own self-contained answer. Add FAQPage and LegalService schema in JSON-LD, and keep the schema matched to the visible text. This is the format Claude extracts most often, and it ports to ChatGPT, Perplexity, and Google AI Mode at the same time.
Third, build the off-site corroboration Claude checks for. Populate and align your profiles on Avvo, Martindale-Hubbell, Justia, your state bar, and Google Business Profile so every detail matches. Earn editorial mentions in legal and local press. The more independent, credible domains confirm your firm, the more confident Claude is to recommend you by name.
Frequently asked questions
Does Claude actually recommend specific law firms by name?
Yes, when a query has local or commercial intent and triggers a live web search, Claude names firms it finds in its results and links to them. For purely definitional questions it answers from training data and usually names no one. The case-driving queries are the ones that trigger a search, so that is where visibility matters most.
Which search engine does Claude use to find law firms?
Claude searches through Brave Search, confirmed by Brave’s place on Anthropic’s subprocessor list and an 86.7% overlap between Claude’s citations and Brave’s top non-sponsored results. This is different from Google AI Mode, which uses Google’s index, so a page can rank on Google and still be invisible to Claude until Brave indexes it.
Why is my law firm not showing up in Claude?
The two most common reasons are that your pages are not yet in Brave’s index, which requires visits from at least 20 unique Brave users with data sharing enabled, and that your pages do not answer the specific question in a clear, structured, first-paragraph block. Inconsistent firm details across directories is the third common cause.
How is getting cited by Claude different from ranking on Google?
Google ranking is about position on a results page. Claude citation is about being one of roughly 10 sources it pulls, surviving its quality and trust assessment, and getting named in a synthesized answer. Claude rewards structure, traceable first-party claims, entity consistency, and corroboration across multiple trusted domains more heavily than raw keyword ranking.
How long does it take to get cited by Claude after fixing my pages?
Most of the lag is Brave indexing. Once Brave picks up a restructured page and your off-site profiles align, firms typically start appearing in Claude answers within four to eight weeks. The same changes usually lift visibility on Perplexity and Google AI Mode in the same window.
Where to start
Claude is the AI engine most law firms have not checked, which makes it the cheapest place to gain ground right now. The work is concrete: get into Brave’s index, restructure your top pages into clean question-and-answer blocks with matching schema, and align the directory and press signals Claude uses to confirm you are real.
If you want to see where your firm stands across Claude, ChatGPT, Perplexity, and Google AI Mode today, run our AI visibility check or estimate the case value you are leaving on the table. Both take a few minutes and show you exactly which moves close the gap.
Tagged