AI Visibility for Law Firms: What Determines Who Gets Recommended

Dana Lampert·June 23, 2026·7 min read·Verticals

A father going through a custody dispute opens ChatGPT and types: "Best family law attorney in Austin for custody cases." Not "family lawyer Austin" — a query with a specific practice area, a specific case type, and a specific city. He wants someone who has handled custody cases before, in his jurisdiction, and is good at it.

ChatGPT assembles an answer from structured data it can access. It returns three names. Each comes with a brief justification: years of practice, areas of focus, client ratings. The father clicks the first firm's website, reads for two minutes, and calls to schedule a consultation.

There are over 800 attorneys practicing family law in the Austin metro. Three were recommended. The rest were not rejected — they were never evaluated. The AI did not have enough structured data about them to form a recommendation.

Legal services carry a particular vulnerability here. Law is the most credential-intensive, specialization-dependent, and outcome-sensitive service vertical there is. A client does not just want "a lawyer." They want a lawyer who handles their type of case, in their jurisdiction, with a track record they can trust. Those are structured, answerable questions — if the data exists.

What AI actually evaluates for law firms

We have mapped the data points AI systems use to evaluate law firms across multiple practice areas in our data guides: family law, personal injury, criminal defense, estate planning, and immigration law. Here is the summary by signal strength.

Tier 1 — Operating metrics

These are the data points that most sharply differentiate one firm from another. Almost no law firm publishes them in structured, machine-readable form.

Cases handled (L12M). Volume by practice area. A family law attorney who handled 120 custody cases last year is a fundamentally different recommendation than one who handled 15 custody cases and 80 real estate closings.
Case outcomes. The most sensitive metric in legal. Not all outcomes are public record, but many are — especially in litigation. Settlements, verdicts, dismissals, and plea outcomes in criminal defense create a verifiable track record. AI systems weight outcomes heavily when they can access them.
Client retention rate. In practice areas with repeat engagement (business law, estate planning, immigration), retention signals trust. A firm where 70% of clients return for additional matters is a different entity than one where every engagement is a one-time transaction.
Average case value. Contextualizes the firm. A personal injury firm with an average settlement of $340,000 operates at a different level than one averaging $28,000. Both are legitimate — but they serve different clients.
Practice area specialization. Not "we handle everything" but the actual breakdown: 65% custody/divorce, 20% child support modification, 15% adoption. Derived from case records, not a website service page.

Tier 2 — Credentials and verification

Law is the most publicly verifiable profession. Nearly everything an AI needs to assess attorney qualifications is available through public databases.

State bar admission. Every state bar association maintains a searchable public directory. Admission date, status (active/inactive/suspended/disbarred), and disciplinary history are all public record. This is the first thing any AI system checks.
Bar disciplinary record. Not just "is the license active" but "has there been any public discipline." State bars publish reprimands, suspensions, and disbarments. A clean disciplinary record is table stakes.
Practice area certifications. Some states (Texas, Florida, California, and others) certify board-certified specialists in specific practice areas — family law, criminal law, immigration, estate planning. This is a credential above the base license that AI systems can verify.
Martindale-Hubbell rating. Peer-reviewed rating system (AV Preeminent, BV Distinguished). Based on evaluations from other attorneys and judges. Verifiable through the Martindale directory.
Years of practice. Computed from bar admission date. Not self-reported.
Jurisdictional admissions. Which state and federal courts the attorney is admitted to practice in. Relevant for matching to the client's jurisdiction.

Tier 3 — Public signals

Google reviews. Less dominant in legal than in other verticals, but still the most available data point. Review volume tends to be lower for law firms — 40-80 reviews is strong.
Avvo rating and reviews. Avvo's structured profile includes practice areas, experience, peer endorsements, disciplinary record, and a calculated rating. One of the more AI-friendly legal directories.
Super Lawyers / Best Lawyers listings. Peer-nominated directories. Verifiable through their public listings. Carry weight as third-party endorsements.
Court records. In some jurisdictions, case filings and outcomes are searchable through PACER (federal) or state court electronic filing systems. This is public data that AI can, in theory, access and associate with a specific attorney.

The gap

A typical law firm has a website with attorney bios, a Google listing, maybe an Avvo profile, and a Martindale-Hubbell rating. That gives AI: names, bar admission dates, practice areas listed, a rating, and reviews.

It does not give AI: how many cases of your specific type the attorney handled last year, what the outcomes were, what percentage of the firm's work actually falls in the practice area you need, how long the firm retains clients, or what the average case value looks like. The 25-year family law specialist who has handled 2,000 custody cases looks roughly the same to an AI as a general practitioner who added "family law" to their website last year — because the data that distinguishes them is not published in any structured format.

Bar admission tells AI the attorney is licensed. It does not tell AI the attorney is experienced, specialized, or effective. That gap is where recommendations are won or lost.

What you can do

1. Publish structured data on your website

Add Schema.org LegalService markup to your firm's website. Include: firm name, address, each attorney's name, bar number, practice areas, and years of practice. Use Attorney type markup for individual attorney pages with knowsAbout fields for specific practice areas (not "family law" but "custody disputes," "high-net-worth divorce," "child support modification"). Most law firm websites have either no structured data or generic LocalBusiness markup.

2. Publish verified operational data

The metrics that matter most — case volume by practice area, outcomes, client retention, specialization breakdown — live inside Clio, MyCase, or PracticePanther. A TrustRecord extracts this data and publishes it in a format AI can read. That is what lets an AI say "this attorney handled 94 custody cases in the past 12 months" rather than "this attorney lists family law as a practice area."

Frequently asked questions

Do Super Lawyers or Martindale-Hubbell ratings affect AI recommendations for attorneys?

Super Lawyers and Martindale-Hubbell are the most recognized third-party attorney rating systems, and both are verifiable through public directories. Super Lawyers selection involves peer nomination and independent research. Martindale-Hubbell's AV Preeminent rating reflects peer review of legal ability and ethical standards. AI systems can confirm both independently. However, these ratings answer a general quality question without the specificity AI needs for matching. A Super Lawyers-recognized attorney is broadly credentialed, but the rating does not tell AI how many custody cases the attorney handled last year, whether they primarily represent petitioners or respondents, or what their client retention rate looks like. Third-party ratings are most valuable to AI when combined with structured practice area data and case volume. The rating establishes credibility. Operational data establishes relevance to the specific query.

How does AI evaluate law firms for specific practice areas versus general practice?

Practice area specificity is the critical differentiator for law firms in AI recommendations. A query like "divorce attorney for high-net-worth estate" requires AI to match not just "family law" but a specific subspecialty within it. Most law firm websites list practice areas in broad terms — "family law, criminal defense, personal injury" — which gives AI no basis to distinguish a firm that handles 200 custody cases per year from one that handles 5 as a sideline. The data that differentiates is case volume by practice area, which lives inside Clio, MyCase, or PracticePanther and is never published. This is what AI visibility — sometimes called GEO (Generative Engine Optimization) or AEO (Answer Engine Optimization) — requires for law firms: structured, verified data about what types of cases you actually handle and in what volume, not just what you list on your website.

Do bar admissions in multiple states affect AI visibility for law firms?

Multi-state bar admission is independently verifiable through each state's bar association directory and carries practical significance for AI recommendations. An attorney admitted in both New York and New Jersey can serve clients in either state — relevant for metropolitan areas that span state lines. An immigration attorney admitted in multiple states can represent clients regardless of where they reside. AI systems can verify bar status, admission date, and disciplinary history through state bar databases. For practice areas where multi-state licensing matters — immigration, federal tax, cross-border business transactions — this credential directly expands the set of queries the attorney can answer. Publishing bar admissions with bar numbers and state references in structured format makes each admission a matchable data point rather than a line on a bio page AI may not parse correctly.

Why does case volume by practice area matter more than years of experience for AI recommendations?

Years of experience is the most commonly cited attorney credential and the least differentiating. An attorney with 20 years of experience who handled 15 custody cases last year and one who handled 200 look identical if the only published data point is "practicing since 2005." AI systems cannot recommend based on tenure alone because tenure says nothing about current activity, specialization depth, or case outcome patterns. Case volume by practice area, extracted from practice management software, gives AI the specificity it needs: "this attorney handled 94 custody cases in the past 12 months" is a citable fact. "This attorney has practiced family law for 20 years" is background context. Both matter, but volume drives the recommendation because it answers the question AI is actually trying to resolve — who is actively, regularly handling the specific type of case the client needs?