The short answer
Law firms should measure AI visibility at the intersection of a client situation, practice area, jurisdiction, matter stage, and eligible lawyer or office. Start with questions a prospective client might ask before choosing counsel, then preserve the complete answer, recommendation role, sources, and every material claim about the firm, lawyers, admissions, locations, experience, fees, and results. Verify those claims against current firm records and the rules and licensing sources that apply in the relevant jurisdiction. Report discovery and factual accuracy separately, with exact denominators, because a visible firm can still be connected to the wrong practice, lawyer, office, or jurisdiction. This is a marketing measurement workflow, not legal or professional-responsibility advice.
Editorial method · evidence boundaries
This guide proposes a review method. Worked examples and pilot sizes are illustrations unless a dated study is explicitly identified. Provider documentation supports only the behavior and product surface it describes; it does not prove that an AIBL workflow improves recommendations or business outcomes.
This is a proposed sector measurement framework. The saved AIBL audit does not establish sector-wide performance, customer outcomes or the effectiveness of this workflow. Domain references support their stated facts, not an AEO benchmark.
Historical AIBL records have limitations: completed jobs can contain truncated answers, and native citation metadata was not retained uniformly. Verify answer usability and source availability before scoring or comparing results. Missing saved metadata is not proof that a provider supplied no sources.
What businesses notice
Common signs of the problem
- The firm appears in a broad answer, but an unrelated office or practice group is presented for the client’s actual need.
- AI answers imply that a lawyer is admitted, certified, available, or experienced in a jurisdiction without current supporting evidence.
- Matter results, rankings, testimonials, fee arrangements, or “specialist” language appear without the qualifications needed to interpret them.
- One firm-wide visibility percentage pools different practice areas, jurisdictions, client types, branded fact checks, provider failures, and ineligible recommendations.
01
Define the prospective-client decision first
This guide is for law-firm marketing, business-development, knowledge-management, web, risk, and practice leaders. The unit of analysis is a defined client decision—not a generic mention of the firm name.
Describe the client situation without inventing a case
Use a realistic but non-confidential scenario with a client type, problem category, matter stage, location, and decision constraint. Avoid entering client facts, privileged material, personally identifying information, or a request for legal advice. The question should test discovery or representation of public facts, not ask an AI system to resolve a live matter.
Resolve the practice area precisely
Map the public practice name to the work the firm actually performs. Keep adjacent areas separate when eligibility, forum, client type, or evidence differs—for example, individual employment claims, management-side employment counseling, executive compensation, and labor relations should not be treated as interchangeable.
Name the jurisdictional context
Record the governing or likely forum only when it is material and supportable: country, state, province, court, agency, or cross-border context. Location is part of the decision cell, not a filter added after collection.
Use the geographic measurement method →Choose one decision stage
Separate open discovery, shortlist building, named-firm comparison, lawyer verification, and contact preparation. A correct answer to “Is this lawyer admitted in New York?” does not prove that the lawyer or firm would enter an unaided shortlist for a New York matter.
02
Build a practice-area decision matrix
Cross only dimensions that could change whether a recommendation is useful or accurate. The matrix should represent real demand and firm capability, not every city-and-keyword permutation the team can generate.
Stratify by practice and client need
Sample the practices and client situations that matter to the firm’s strategy, then document which teams are genuinely eligible for each cell. Include different client types only when their needs or the firm’s offering differ. Do not count visibility for work the firm does not seek or cannot handle.
Add jurisdiction and forum deliberately
Create separate cells when admission, local procedure, office coverage, governing law, or forum experience could affect the choice. ABA Model Rule 5.5 addresses unauthorized and multijurisdictional practice, including representations about admission, but actual adopted rules and exceptions vary. The firm’s qualified reviewer should decide the applicable boundary.
Read ABA Model Rule 5.5 ↗Represent the client’s decision stage
For each practice-jurisdiction pair, choose the meaningful stages: identify a category of counsel, find firms with a relevant capability, compare named firms, verify a lawyer or credential, or locate the right office and contact. Remove ceremonial cells that cannot change a decision.
Keep a stable core and rotating edge
Freeze high-consequence cells for trend comparison. Rotate emerging issues, newly launched practices, changing offices, new lawyer admissions, and under-tested client segments through an exploratory set. Version every change to wording, eligibility, market, provider, and source record.
Design a coverage-led question portfolio →03
Separate discovery, comparison, and verification
Supplying a firm or lawyer name changes what the observation can prove. Keep question families apart so easy branded checks do not inflate unaided visibility.
Unaided practice discovery
Ask which firms could plausibly help a defined client type with a defined matter and jurisdiction without naming the firm. Record eligible-firm inclusion, recommendation role, rationale, caveats, lawyers or offices named, and sources. An ineligible recommendation is an accuracy problem, not a visibility win.
Aided firm comparison
When prospective clients commonly compare named firms, use a fixed eligible set and label the observation as aided. Capture the comparison dimensions, stated differences, material claims, caveats, and citations. Test candidate order separately from the core discovery measure.
Use the matched prompt-context protocol →Lawyer and office verification
Use branded questions to check a named lawyer’s public role, practice, office, admissions, certification, languages, and contact path, or an office’s current location and services. These cells measure the accuracy of public facts after awareness; they do not establish market preference.
General legal information is a separate test
An answer explaining a legal concept may cite the firm without recommending it, and it may give inaccurate or incomplete information despite naming the firm correctly. Track citation, legal-information quality, and firm recommendation separately. Have qualified lawyers assess consequential legal statements; the marketing team should not score substantive law from intuition.
04
Create a firm, lawyer, and claim ledger
Every observed statement should resolve to the right entity, time, jurisdiction, and supporting record. Similar firm names, lateral moves, office changes, and overlapping practice labels make entity resolution essential.
Preserve the complete observation
Store the exact question, full response, provider, visible product or model label, timestamp, search state, conversation state, jurisdiction, citations, attempt status, and exclusion reason. Keep provider failures and blocked responses outside brand-omission denominators.
Resolve firm and lawyer identity
Map the answer to the exact legal entity or public firm brand, lawyer, title, office, practice group, canonical profile, and applicable effective date. Flag ambiguous names and former lawyers instead of assigning the answer to the most favorable match.
Use explicit factual states
Classify each material claim as verified, contradicted, outdated, ambiguous, unsupported, not applicable, or not checked. Save the source URL, record owner, effective date, jurisdiction, checked-at time, and exact answer text. Preserve qualifications instead of reducing a nuanced claim to a binary score.
Separate presence, fit, and accuracy
Code whether the firm appeared, whether it was eligible for the defined situation, how it was positioned, and whether material statements were supportable. A prominent recommendation tied to the wrong jurisdiction or a departed lawyer is not a successful observation.
05
Review material claims against controlling sources
Legal-services communications are regulated by jurisdiction. The ABA Model Rules are influential models, not a substitute for the rules, opinions, statutes, and decisions that actually govern a firm or lawyer.
Use the applicable rule set
Identify the jurisdictions connected to the lawyer, office, audience, and communication, then have qualified counsel determine which rules apply. The ABA publishes comparison charts showing that jurisdictions do not uniformly adopt the Model Rules. Do not treat one national checklist as approval for every market.
Review ABA jurisdictional comparison charts ↗Check for false or misleading descriptions
ABA Model Rule 7.1 prohibits false or misleading communications about a lawyer or the lawyer’s services, including material misrepresentations or omissions. Use that model principle as a review prompt for AI statements about experience, outcomes, availability, rankings, clients, and services, then apply the firm’s actual governing rule.
Read ABA Model Rule 7.1 ↗Verify specialization and responsible-party claims
Current ABA Model Rule 7.2 permits communications through any media but places conditions on stating or implying specialist certification and calls for the name and contact information of a responsible lawyer or firm. State rules can differ. Record the exact AI wording, certification body if named, and current firm-approved language rather than substituting labels such as “expert” or “specialist.”
Read ABA Model Rule 7.2 ↗Verify admissions and status directly
Check bar admission, status, certification, court admission, office, and title against current licensing-authority records and firm records at the scope required for the claim. A biography, directory, old press release, or AI citation may be stale. Record the source and check date; do not infer authority to handle a matter from a generic firm location.
06
Interpret results by practice, jurisdiction, and consequence
A firm-wide average can hide the precise mismatch that matters to a prospective client or creates review risk. Keep the matrix slices visible before producing any executive summary.
Publish exact denominators
Show planned, attempted, completed, failed, excluded, and not-applicable observations by practice, client situation, stage, jurisdiction, provider, question family, lawyer, and office. Say “included in 3 of 6 completed unaided discovery cells for the tested Illinois employment scenarios,” not “visible in 50% of legal AI.”
Find discovery-to-verification breaks
Identify practices that enter a shortlist but fail when the client verifies jurisdiction, lawyer identity, office, certification, or scope. Also find accurate and eligible practices that remain absent from relevant unaided discovery. Those are different problems with different owners.
Prioritize by client and communication risk
Escalate wrong jurisdiction, admission, lawyer identity, certification, availability, fee, deadline, or outcome claims before harmless wording differences. Rank findings by likely client consequence, applicable review risk, recurrence, evidence quality, and the owner who can correct the authoritative source.
Repeat bounded cells before calling a trend
Run the same question-provider-jurisdiction cell in a fresh context under a predeclared repeat rule. Keep ordinary run variance separate from prompt edits, lawyer or rule changes, news events, provider disagreement, and evidence updates.
Use the repeat-run stability protocol →07
Run a bounded pilot and assign owners
This fictional example demonstrates scope and arithmetic only. It is not a client result, legal-market benchmark, ethics opinion, claim about any provider, or recommended minimum for every firm.
Choose twelve decision cells
Fictional Harbor & Pine selects three priority practices, two prospective-client stages for each, and two materially different jurisdictions or client contexts. Every row names the eligible firm team, authoritative sources, material claims to review, and responsible marketing, practice, web, or risk owner.
Make observation volume explicit
Twelve questions across ChatGPT and Gemini with two fresh-context repeats create 48 planned observations: 12 × 2 × 2. Those are answer attempts, not 48 unique client needs, and they do not represent the legal market. Failures, exclusions, and ineligible recommendations stay visible.
Route the finding before changing content
Send lawyer and practice facts to profile owners, jurisdiction and professional-responsibility questions to qualified firm counsel, office facts to operations, and discovery gaps to marketing or web owners. Write the observed answer, verified truth, source, hypothesis, proposed change, approval path, and retest rule.
Turn findings into an action plan →Write a bounded conclusion
The report can state which tested firms or practices appeared, how they were positioned, and which observed claims were supported or contradicted for the documented cells. It cannot claim market share, professional quality, likely case outcomes, compliance, or provider-wide treatment beyond those observations.
08
Start with one non-confidential client question
The free AI visibility snapshot provides one inspectable ChatGPT answer and one Gemini answer for the same buyer-style question. For a law firm, that is one discovery cell—not legal advice, a complete advertising review, or proof of eligibility for a matter.
Choose a defined but safe question
Ask about one public practice need, client type, decision stage, and jurisdiction without naming the firm and without including confidential, privileged, or personal facts. Review eligible-firm inclusion, recommendation role, rationale, lawyers or offices named, cautions, and sources.
Run a free AI visibility snapshot →Verify before reusing the answer
Check every material firm, lawyer, admission, certification, office, service, fee, result, testimonial, ranking, and legal statement against current authoritative sources and the applicable review process. Do not reuse an unchecked AI answer in marketing or client communications.
Expand only when another cell matters
Add a question when another practice, client situation, stage, jurisdiction, forum, lawyer, or office could change eligibility or action. Keep cosmetic paraphrases and candidate-order tests outside the core matrix so the measurement remains interpretable.
Use the snapshot as an evidence baseline
If the two answers reveal a consequential omission or unsupported claim, preserve them as dated evidence and build the smallest matrix needed to diagnose it. AI Brand Lens measures observable answers and sources; it does not provide legal advice or guarantee recommendations, clients, compliance, or outcomes.
Continue the research
Related AI visibility guides
Primary sources
Primary documentation used for this guide
AI products, search behavior, and platform policies change. Check these maintained first-party sources before making technical decisions.
- Model Rules of Professional Conduct ↗
American Bar Association
The ABA explains that the Model Rules serve as models for most jurisdictions. The guide treats them as an issue-spotting framework, not controlling law, and directs firms to the rules and authorities that actually apply.
- Rule 7.1: Communications Concerning a Lawyer’s Services ↗
American Bar Association
The current Model Rule prohibits false or misleading communications about a lawyer or the lawyer’s services, including material misrepresentations and omissions. Jurisdiction-specific adoption and interpretation must be checked separately.
- Rule 7.2: Communications Concerning a Lawyer’s Services — Specific Rules ↗
American Bar Association
The current Model Rule covers communications through media, recommendation payments, specialist-certification language, and responsible-party identification. The guide uses these as material claim categories and does not assume uniform state adoption.
- Rule 5.5: Unauthorized Practice of Law; Multijurisdictional Practice of Law ↗
American Bar Association
The Model Rule addresses practice across jurisdictions and representations about admission. It supports recording jurisdiction and admission precisely, while the guide leaves applicability and exceptions to qualified review under controlling authority.
- Jurisdictional Rules Comparison Charts ↗
American Bar Association
ABA-maintained comparison resources showing variation between Model Rules and jurisdictional rules. They support the guide’s warning against treating a national model as the controlling rule in every jurisdiction.
- Evaluation best practices ↗
OpenAI Platform Documentation
First-party guidance on task-specific objectives, representative test cases, explicit metrics, logging, human judgment, and continuous evaluation. The article adapts those principles to observable legal-services answers without implying control over consumer AI products.
Measure your visibility
Turn the questions in this guide into an evidence-backed baseline.
Get AI Visibility ScoreTalk to us
Have a visibility question?
Tell us what your team is trying to measure or improve.
Contact AI Brand Lens →Keep learning
Explore every guide.
Browse practical answers about AI visibility, competitors, and measurement.
View all guides →