An AI visibility audit law firm owners can run themselves is a repeatable test of whether ChatGPT, Gemini, Perplexity and Google’s AI features name your firm, cite your website and describe you correctly when a prospective client asks a family law question. You can run one yourself. Confirm that AI crawlers can reach your site, ask 30 realistic questions across three platforms, log what comes back, check that your firm’s details match across directories, then score the result and fix the biggest gap first. Buy a monitoring tool only after you’ve done it once by hand.
Family law makes this harder than most practice areas, and the reason is the query. Nobody asks an AI for “a lawyer.” They describe a situation: a move across state lines the other parent opposes, a spouse who may be hiding income. Google says its AI Overviews and AI Mode may use “query fan-out,” running several related searches across subtopics before composing an answer. So your visibility depends on whether your pages answer the sub-questions inside that situation, the same groundwork any serious program of SEO for family lawyers covers. The audit shows where the AI layer departs from it.
How do I check if ChatGPT recommends my law firm?
Open a logged-out ChatGPT session and ask what a prospective client would, such as “Who are the best family law attorneys in [county], [state]?” Record whether your firm is named, linked and described correctly. Repeat key prompts three times in new chats, because answers can differ run to run. One answer proves nothing; the pattern across 30 prompts does.
Thirty is enough to see a pattern without burying a paralegal. Write six prompts for each bucket below, using your real county, city and attorney names.
| Bucket | What it tests | Example prompt |
| Branded | Whether the firm reads as one clear entity | “Is [Firm Name] a good family law firm in [City]?” |
| Shortlist | Whether you get recommended | “Who are the best family law attorneys in [County], [State]?” |
| Situation | Whether your pages answer the real problem | “My ex wants to move out of [State] with our child. Who should I talk to?” |
| Process | Whether your content gets cited as a source | “How does a contested custody case move through court in [State]?” |
| Cost and fit | Whether intake and fee information can be found | “How do family law retainers work in [City], and what should I ask before hiring?” |
Run them in ChatGPT with search on, in Gemini and in Perplexity, each in a logged-out window and a fresh chat. That’s 90 answers. Log six things for each: the prompt, the platform, whether you were named, whether your own URL was cited, whether the details were right, and who was named instead.
Then take the ten prompts worth the most money to the firm and run them twice more. Appear in one run out of three and you have an unstable signal, not a ranking.
The AI visibility audit law firm owners can score in four passes
Do them in order. A failure in Pass 1 makes everything after it irrelevant.
Pass 1: Can the systems reach your site?
Google’s documentation says a page must be indexed and eligible to appear with a snippet before it can be a supporting link in AI Overviews or AI Mode. It also says no special optimization is required beyond normal SEO fundamentals. Start in Search Console. Confirm your practice area pages and attorney bios are indexed, then hunt for noindex tags or snippet restrictions that someone added years ago and nobody remembers.
ChatGPT works differently. OpenAI runs separate crawlers, and its documentation says each robots.txt setting is independent: OAI-SearchBot surfaces sites in ChatGPT search, while GPTBot concerns model training. A firm can disallow GPTBot and still appear in ChatGPT search if OAI-SearchBot is allowed. Plenty of agency checklists tell you to look for GPTBot first. For visibility, that’s the wrong crawler.
Open yourdomain.com/robots.txt and look for blanket disallow rules and any rule naming OAI-SearchBot. OpenAI also recommends allowing its published IP ranges, so check your firewall, CDN and WordPress security plugin; those block bots without ever touching robots.txt. After a change, allow about a day. OpenAI says its systems take roughly 24 hours to adjust.
Pass 2: What does the test show?
Count the prompts where you were named in at least one run. Then count how often your three closest local competitors appear. Their numbers are your benchmark. A target percentage from a vendor’s blog describes that vendor’s clients, not your market.
A wrong detail outranks everything else on the sheet: a dead phone number, a departed attorney, a practice area you don’t handle. Fix those first, at the source the AI used.
Pass 3: Do your profiles agree with each other?
AI systems assemble a firm from many places. Pull up your Google Business Profile, your state bar directory entry, Avvo, Justia, FindLaw and LinkedIn, then compare six fields: firm name spelling, each attorney’s name format, address and suite, phone, practice areas and bar admissions. Mismatches surface after a move, a merger or a name change.
Score a mismatch only if it’s material. A missing “LLC” is noise. A phone number from the old office is not. And don’t buy a new listing because an AI cited it once. Check whether that source shows up across runs first.
Pass 4: The GEO audit a family law firm’s pages need
Take the five pages that should win your prompts and read them as a stranger would. Does the first paragraph answer a specific question? Is the state or county named? Is a real attorney attached, with a bar admission, rather than “our team”? Are procedural claims tied to a court or statute page?
Schema comes last. Google says nothing special is required for its AI features, so markup is not a switch. Use LegalService or Attorney markup (check schema.org for the current definitions) to state facts already visible on the page, and validate it with Google’s Rich Results Test. Markup that contradicts the page is worse than none.
For measurement, Search Console counts AI feature traffic inside overall Search totals. Google also announced a generative AI performance report in June 2026; check whether your property has it.
What should the scores make me fix first?
Read the sheet by pattern, not by total. There is no industry benchmark here worth trusting, so the table below ties each pattern to a likely cause and a first move.
| Pattern on your sheet | Likely cause | First fix | Skip for now |
| Never named, own site never cited, Pass 1 has failures | Crawlers or indexing blocked | Remove the block, then retest | Content rewrites |
| Named only on branded prompts | Entity is clear, topical coverage is thin | Build pages for the situation prompts you lost | New directory listings |
| Named, but details wrong | Outdated or conflicting profiles | Correct Google Business Profile, state bar entry and your own bios | Schema work |
| Competitors named, you cited only on process prompts | Content is good, third-party validation is thin | Review and profile consistency | More blog posts |
| Named in one run of three | Unstable signal | Retest monthly, change nothing yet | Any spend based on one run |
What this audit can’t tell you
A single AI visibility audit law firm owners run is a snapshot. Results shift by platform, by day and by who is asking, so the first audit is a baseline and a second one a month after your fixes is the first real comparison. The audit says nothing about search volume, case value or whether a mention turns into a consultation; your intake log answers that.
Nobody controls what an AI system says. Treat anyone who guarantees a ChatGPT recommendation, including us, as a red flag.
Do AI recommendations create an advertising-rule problem?
Not by themselves, but whatever you publish to influence them is still a communication about your services. California’s Rule 7.1 says a lawyer shall not make a false or misleading communication about the lawyer or the lawyer’s services, and the District of Columbia’s version adds a bar on assertions that cannot be substantiated. Wording and numbering differ by state. Before you rewrite bios or profiles for AI visibility, read your own state bar’s rule and have ethics counsel look at the draft.
FAQ
What is an AI visibility audit for a law firm?
A structured test of whether AI systems such as ChatGPT, Gemini, Perplexity and Google’s AI features name your firm, cite your site and describe it accurately for the questions prospective clients ask. It combines a technical access check, a prompt test, a directory consistency review and a page-level review.
How often should I rerun it?
Run a second audit about a month after your first fixes so you compare like with like, then quarterly, and again after a site migration, a rebrand or an attorney change. Compare patterns across prompts, not single answers.
Do I need to allow GPTBot to appear in ChatGPT search?
No. OpenAI documents OAI-SearchBot as the crawler for ChatGPT search results and GPTBot as a separate setting tied to model training. Allow OAI-SearchBot, make a separate decision about GPTBot, and check that firewall rules don’t block OpenAI’s published IP ranges.
Does ranking well on Google guarantee AI recommendations?
No. Google says AI Overviews and AI Mode may run several related searches across subtopics before answering, so your pages are tested against the sub-questions, not only the main query. A page that ranks for the head term can still go uncited.
Can an agency guarantee that ChatGPT will recommend my firm?
No. Nobody controls what an AI system outputs. Ask any vendor making the promise how they would measure it, and treat a guarantee as a reason to walk away.
Where do I see Google’s AI feature traffic?
In Search Console, where traffic from AI features is counted inside overall Search totals. Google announced a dedicated generative AI performance report in June 2026; check whether your property has it.
