At 2am, someone asks ChatGPT if they need a DUI lawyer. They call whoever gets named.
Not the firm with the billboards, or the biggest Google Ads budget, or the best trial record — the firm the AI names. Hoss measures whether that's you, fixes the reasons it isn't, and re-measures every month.
AI visibility for a law firm is whether ChatGPT, Gemini, Perplexity, and Claude name your firm when potential clients ask legal questions and ask who to hire. For legal queries the engines lean heavily on established directories — Avvo, Super Lawyers, Justia — plus Reddit threads, practice-area content with real depth, and review signals. It can be measured, and the gaps can be fixed.
Why does the AI answer decide who gets the call?
Because legal questions are asked at the worst moments — after the arrest, after the accident, after the letter arrives — and the person asking takes the first credible answer. AI assistants don't return ten blue links. They return two or three firm names, stated with confidence.
A potential client asks AI
"Do I need a DUI lawyer?" "Best personal injury attorney near me." "Which firm do people actually recommend?"
AI names a short list
Composed from legal directories, community threads, and practice-area pages the engines can read and corroborate.
Absence is invisible
The client calls one of the named firms. High-value matters route to competitors before you know the question was asked.
What determines which law firms get named?
Legal is one of the most directory-dominated verticals in AI search. One legal-SEO analysis found 87% of ChatGPT citations come from pages already in the top 10 search results, with established legal directories — Avvo, Super Lawyers, Justia — dominating, and roughly half of Perplexity's citations tracing to Reddit [1].
Directories, Reddit, reviews
Where the engines actually look for legal answers.
- Legal directories — Avvo, Super Lawyers, Justia profiles are what AI cites for lawyer-selection queries; incomplete or inconsistent profiles forfeit the citation [1].
- Reddit — disproportionately influential: a Semrush study (June 2025) of 150,000+ LLM citations found 40.1% pointed to Reddit, ahead of Wikipedia at 26.3% [2]. Legal-marketing guidance treats consistent, genuinely helpful Reddit participation — not spam — as compounding work [3].
- Review signals — volume, recency, and consistency across Google and the directories, so the engines see one coherent, corroborated firm.
Practice-area depth, machine-readable
What the engines read on your own site.
- Practice-area pages with depth — a page per matter type that answers the questions clients actually ask, in their language. Reddit threads are where that real client language lives [4].
- Extractable answers — direct answers under question headings, positioned where a retrieval system can lift them cleanly.
- Structured data and crawler access — schema that tells machines what the firm is and where it practices, and robots.txt/firewall rules that let AI crawlers read any of it.
Which questions do potential clients actually ask?
Two kinds: crisis questions, asked in the moment, and selection questions, asked once they've decided to hire. Both end in a short list of names.
The crisis questions matter more than they look: the firm whose content answers "do I need a lawyer for this" is the firm the engine already trusts when the "who should I hire" question follows. You can run this exact check on your own firm, free, in about fifteen minutes — here's the complete method.
Measure. Fix. Prove it monthly.
The same loop we run for every client, instantiated for your firm and practice areas. No black box: the full methodology is published, and you can see a sample report before you spend anything.
The teardown
We build a prompt suite from real client questions — crisis and selection, per practice area, in your market — and run it across the engines, multiple times per prompt. You get where you're named, which firms win instead, which directories and threads the engines cite, and what AI gets factually wrong about your firm.
The engagement
A scoped fix list from your teardown: directory profiles completed and made consistent, practice-area pages rebuilt around real client questions with extractable answers, schema and entity work, and the crawler and firewall repairs that let engines read all of it.
The scoreboard
Every month, the same prompt suite re-runs and you see mention rate, recommendation rate, and citation share — against the other firms in your market, by name. You see exactly what moved, and what didn't.
No one can guarantee your firm a spot in AI answers. Anyone who promises otherwise is selling something else.
Even legal-marketing practitioners writing about this work put it plainly: no agency can honestly guarantee AI visibility outcomes [4]. What an honest one can do is measure, systematically raise the probability you're retrieved and recommended, and publish the scoreboard every month.
What managing partners ask before they book.
Do potential clients really use ChatGPT to find a lawyer?
Yes — legal questions that once went to Google at 2am now go to AI assistants, and the answer comes back as a short list of firms composed from directories, Reddit threads, and practice-area content. The firm that gets named gets the call.
What determines which firms AI names?
For legal queries: established directories (Avvo, Super Lawyers, Justia) dominate citations, 87% of ChatGPT's citations come from top-10 search results, and Reddit carries outsized weight — about half of Perplexity's citations trace to it [1]. Practice-area depth and review consistency complete the picture.
Can you guarantee we'll be recommended?
No — no agency can honestly guarantee AI citations, and anyone who promises otherwise is selling something else. We measure, fix the causes we find, and re-measure monthly so you can see exactly what moved.
Can I check where my firm stands without hiring anyone?
Yes. Run the questions your clients ask through the engines in fresh chats, two or three times each, and record which firms get named and which sources are cited. We published the complete free method.
What about attorney advertising rules?
They apply, and you should treat any marketing vendor's work as your responsibility under your state bar's rules — because it is. How we work with that reality: nothing is published on your behalf without your firm's review and approval, the content we build is factual practice-area material rather than comparative "best lawyer" claims, and much of the engagement (crawler access, structured data, entity consistency, directory accuracy) isn't advertising copy at all. Your bar counsel has the final word; we build to make that review easy, not to route around it.
Does being named by AI actually produce clients?
Being named is necessary but it isn't a retainer, and we won't pretend the line from mention rate to signed clients is automatic. What we measure — named, recommended, cited, monthly, against competitors — is the part of the funnel we can affect and prove. Connecting it to intake is something we set up with you at the start: matched intake questions ("how did you find us?"), consult-source tracking, and honest reporting when the numbers move but the phone doesn't.
Sources
- AI Vortex — legal AI-search analysis: 87% of ChatGPT citations from top-10 search results; Avvo, Super Lawyers, and Justia dominate legal citations; roughly half of Perplexity's citations trace to Reddit.
- Custom Legal Marketing, citing Semrush (June 2025) — 150,000+ LLM citations analyzed; Reddit 40.1%, Wikipedia 26.3%, YouTube 23.5%.
- Attorney at Work — law-firm Reddit marketing dos and don'ts; 15–30 minutes twice weekly of genuine participation compounds.
- Follow White Rabbit — Reddit as a source of real client questions and language for law firms, and the plain statement that no agency can honestly guarantee AI-visibility outcomes.
Find out what AI tells potential clients about your firm.
A 30-minute call. We'll run real client questions for your practice areas through the engines, live, and show you exactly which firms get named — including whether you need us at all.
Book a callPrefer email? [email protected]