Classic rank tracking has a hundred polished tools. AI visibility has almost none that work well for a single law firm — the engines personalize, answers vary run to run, and there is no public "position one." The answer is not to give up on measurement; it is to run a small, disciplined panel the way pollsters do. An hour a month produces a trend line you can actually manage against.
Build a fixed query panel
Choose fifteen to twenty-five questions that represent real client intent for your firm: five brand queries ("[firm name] reviews"), ten practice-area money queries ("how much does a divorce cost in [province/state]", "do I need a lawyer for a first DUI"), and five research queries a step earlier in the journey. Write them down and do not change them month to month — the value is in the trend, and a moving panel destroys the trend. When you add practice areas, add a new panel section rather than editing the old one.
Score four engines the same way
Each month, run the panel through ChatGPT (with search enabled), Google AI Overviews, Gemini, and Perplexity. For every query, record three things: whether your firm is mentioned in the answer text, whether one of your pages is cited as a source, and which competitors appear. That gives you a mention rate and a citation rate per engine — two numbers per engine, eight numbers total. Log the exact page URL cited when there is one; over time this reveals which of your pages the engines actually trust, which is where you should concentrate updates.
- Mention rate: answers naming your firm ÷ panel size
- Citation rate: answers linking one of your pages ÷ panel size
- Competitor share: which three names appear most across your panel
- Cited-page log: the URLs engines choose, month over month
Why the panel must include losing queries
A tempting mistake is to build the panel from queries you already win — it produces a flattering dashboard and no information. Deliberately include the questions where competitors dominate today, the ones your intake team says clients actually asked before calling someone else. Those rows will read zero for months, and that is their job: the month one flips from absent to mentioned is the clearest evidence you will ever get that a specific piece of content work moved a specific engine. Panels made only of victories cannot show causation; panels with honest losses can.
Wire up the referral data you already have
AI engines now send measurable traffic. In your analytics, segment sessions with referrers from chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com, and annotate your consultation conversions against them. Most firms find AI-referred visitors convert at two to three times their organic rate — they arrive having already read a synthesized comparison in which your firm looked good. That conversion premium is the business case for the whole effort, so track it explicitly rather than letting it hide inside "referral."
The one-hour monthly workflow
Discipline beats tooling here, so make the session mechanical. Block one hour on the first business day of each month. Open a spreadsheet with one row per panel query and four column groups, one per engine; each group holds three cells — mentioned, cited page, competitors named. Work through the panel in order, one engine at a time, pasting the cited URL when there is one. Use a clean browser profile with no account history so personalization does not flatter you: you want to see what a stranger sees, because your future clients are strangers.
Resist the urge to re-run a query that came out badly — a second roll of the dice is not data. Record the miss and move on; if it is real, it will still be there next month, and if it was variance, the trend will absorb it. At the end of the hour, compute the eight rates, write two sentences of interpretation ("Perplexity began citing the fee guide; Gemini still prefers the bar association for cost questions"), and close the file. Twelve of these sessions produce a year-long visibility record that no off-the-shelf tool can currently match for a single firm, at a cost of one billable hour a month.
Read the results like a strategist, not a scoreboard
Variance is normal — the same question can produce different answers an hour apart. Judge direction over quarters, not months. When an engine mentions you but cites a competitor's page for the substantive claim, that is a content gap on a specific topic; write the better page. When you are absent entirely from one engine while healthy on the others, that is usually a technical or crawler-access issue for that engine's bot — check it before writing anything. And when your monthly report shows citations concentrating on two or three pages, protect those pages: keep their facts current, their dates fresh, and their schema intact, because they are doing disproportionate work for the whole domain.
A firm that runs this panel for six months knows more about its AI presence than most agencies can tell it — and every improvement it makes is verifiable in the next month's numbers. That verifiability is the real prize: in a channel this new, the ability to prove what worked is itself a competitive advantage, because you can double down on evidence while competitors guess.
Frequently Asked Questions
Grow your AI SEO practice with AI
Lexscale.ai builds AI search visibility, websites, and intake systems for ai seo firms across North America. Book a free strategy call to see what would move the needle for your practice.
Book a Free Strategy Call →