How to Track Your Firm’s Visibility in AI Answers
The Short Answer
Tracking AI visibility means running a fixed set of client questions through ChatGPT, Gemini, Perplexity, and Google’s AI surfaces on a schedule, then recording whether your firm is mentioned, cited, and recommended over time. You can start free with a spreadsheet and thirty minutes a month, or pay for tools that automate it from about 29 dollars a month up. Either way, the metrics that matter are mention rate, citation share, and share of voice against named competitors. One time checks are nearly worthless because AI answers vary from run to run; trends over months are the only signal worth acting on.
Every personal injury firm knows its Google rankings. Almost none know whether ChatGPT recommends them, whether Gemini quotes their content, or what Perplexity says when someone asks who to call after a truck accident in their city. That gap is where the next few years of client acquisition quietly get decided.
The good news: tracking AI visibility is easier than tracking rankings ever was, because you can see the answer yourself with a free account. The bad news: it is easier to do wrong, because AI answers shift constantly, and a single check can tell you a story that is not true. This guide covers the manual method, the paid tools worth knowing, the metrics that matter, and the mistakes that make firms chase ghosts.
Why One Time Checks Mislead You
Before any method, understand the measurement problem. AI answers are not stable rankings. They are generated fresh each time, and the same question asked twice can name different firms and cite different sources.
The research on this is sobering. A peer reviewed study of 11,500 queries presented at SIGIR in 2026 found that Google AI Overviews returned less consistent sources on repeat runs than organic results did, and that even cosmetic edits to a page cut source agreement by nearly 29 percent. Separate testing by tool vendors found that answers collected through official APIs matched the real consumer interfaces only about a quarter of the time. Translation: the answer you see today, on your device, in your city, is one sample from a moving distribution.
Three habits protect you from bad data. First, always run the same frozen set of questions, word for word, every time. Second, run each question multiple times across multiple days and count how often you appear, not whether you appeared once. Third, log everything with dates, because a month of logs is information and a single screenshot is an anecdote.
The Free Method: A Prompt Set and a Spreadsheet
Every firm can run a credible tracking program for zero dollars. It takes an hour to set up and thirty minutes a month to maintain.
1. Build a prompt set of 15 to 25 questions.
Split it three ways. Recommendation questions: best personal injury lawyer in Los Angeles, who should I call after a car accident in the Valley. Informational questions: how long do I have to file an injury claim in California, what is my slip and fall case worth. Name questions: your firm name, your name plus reviews, your name plus the practice area. Use the client’s words, gathered from intake calls, not your marketing language. Freeze the wording permanently.
2. Run it monthly across the major surfaces.
At minimum: ChatGPT, Perplexity, the Gemini app, and a regular Google search where you note whether an AI Overview appears and who it cites. Run each prompt three times on different days if you can, and log the modal result. Consistency of process matters more than volume.
3. Score every answer the same way.
For each prompt and platform, record four things: was your firm mentioned at all, was your website cited as a source, which competitors appeared instead, and which third party pages were cited. Those last two columns are where the strategy lives, because they tell you exactly who you have to displace and what the machine is reading.
4. Review the log quarterly, act monthly.
Month to month wiggles are noise. Quarter over quarter direction is signal. When a competitor starts appearing where you used to, open their cited pages and ask what changed: fresher dates, better answers, new reviews, a directory you ignored. The log turns AI visibility from a mystery into a maintenance task.
PRIORITY CALLOUT
If you do nothing else from this article: write down fifteen client questions today, run them through ChatGPT and Perplexity tonight, and save the results with the date. That single snapshot becomes the baseline every future improvement is measured against. Firms without a baseline cannot prove progress, so they stop investing. Firms with one compound.
The Paid Tools, Honestly Assessed
When the spreadsheet gets tedious, a category of AI visibility tools automates the running, logging, and scoring. The category is young and the prices move fast, but as of late 2026 the landscape looks like this:
| Tool | Entry Price | What You Get | Best Fit |
|---|---|---|---|
| Otterly AI | From $29/mo | 15 tracked prompts across ChatGPT, Perplexity, AI Overviews, Copilot | Cheapest credible start |
| ZipTie | From ~$26/mo | 30 prompts checked daily, deep citation analysis | Hands on, detail hungry users |
| Peec AI | From $95/mo | 50 prompts, unlimited seats, suggested actions | Small teams sharing access |
| Semrush AI Toolkit | $99/mo | 25 prompts daily next to your existing SEO data | Firms already on Semrush |
| Ahrefs Brand Radar | From $199/mo | Instant data from a database of roughly 500 million prompts | Benchmarking without setup |
| Profound | Enterprise | Deep answer intelligence across up to nine engines | Large firms and agencies |
A few free on ramps are worth using before paying anyone. Ahrefs offers a free AI visibility checker with no signup, and HubSpot has a free AI search grader. Neither replaces tracking, but both make a decent first look.
Two buying cautions. First, check how the tool collects answers: tools reading the real consumer interfaces reflect what your clients actually see, while API based tools can diverge sharply, per the mismatch research above. Second, price the engines you need, not the headline. Several tools charge extra per AI platform, and a cheap plan covering two engines may cost more at your actual coverage than a pricier plan covering six.
The Metrics That Actually Matter
Whatever method you use, reduce the flood of data to five numbers.
Mention rate: across your prompt set, what share of answers name your firm at all. This is your baseline health metric.
Citation rate: how often your own pages are cited as sources. Mentions build reputation; citations build traffic and authority, and the two do not always move together.
Share of voice: of the firms named across your recommendation prompts, what percentage are you. Pick two or three named competitors and track them every cycle. AI visibility is relative; you win by displacing someone.
Sentiment and accuracy: when the machine talks about you, is it right? Wrong practice areas, stale addresses, and confused identities are entity problems, and catching them early is half the value of tracking.
AI referral traffic: in your analytics, segment traffic from chat.openai.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com. The volumes are still small for most firms, but they convert at high intent, and the trend line is the proof your visibility work is reaching actual humans. Set this segment up once and it runs forever; checking it takes two minutes of your monthly review.
One more habit ties all five metrics together: pair every number with the work you shipped that month. Published three new guides, claimed two directory profiles, pushed twenty new reviews? Note it in the log next to the scores. Within two quarters you will have something no tool sells, a cause and effect record of which activities move your firm’s visibility and which ones just keep you busy. That record becomes your budget defense and your strategy at the same time, and it is the first thing any competent marketing partner will ask to see.
The Mistakes That Waste the Effort
Tracking vanity prompts. Checking only your own firm name tells you whether the machine has heard of you, not whether it recommends you to strangers. The money prompts are the category and city questions asked by people who have never heard your name.
Checking once and declaring victory or defeat. Given how much answers vary run to run, a single bad result means almost nothing and a single good one means less. Decide on trends across at least a quarter.
Changing five things at once. If you overhaul your site, your profiles, and your content in the same month, your log cannot tell you which change moved the needle. Sequence the work, keep the prompt set frozen, and let the data attribute the wins.
Tracking without acting. A visibility log that never triggers a page update, a profile claim, or a review push is a very organized way to watch yourself lose. Every review meeting should end with one assigned fix.
This tracking loop, frozen prompts, monthly runs, quarterly reviews, assigned fixes, is the measurement backbone of our AI search optimization work for personal injury lawyers, because in a channel this new, the firms that measure are the only ones that know whether they are winning.
Frequently Asked Questions
Can I track AI visibility for free?
Yes. A frozen set of 15 to 25 client questions, run monthly through ChatGPT, Perplexity, Gemini, and Google, logged in a spreadsheet with dates, covers the fundamentals. It costs thirty minutes a month. Paid tools automate the same loop and add competitor benchmarking.
How often should I check my firm’s AI visibility?
Monthly is the right cadence for most firms. Weekly produces noise you will overreact to; quarterly lets problems fester. Whatever cadence you choose, keep the prompt wording frozen and review trends quarterly rather than reacting to single runs.
Why do I get different answers every time I ask the same question?
Because AI answers are generated fresh each run and vary by device, location, and timing. Research shows source citations in AI Overviews are meaningfully less consistent than organic rankings. Run prompts multiple times over several days and count how often you appear, never whether you appeared once.
What is a good mention rate for a local PI firm?
There is no universal benchmark yet, which is exactly why your own baseline matters. For competitive Los Angeles recommendation prompts, appearing in a third of runs is a strong early position. The goal is steady quarter over quarter growth against the same frozen prompt set.
Do I need an expensive tool like Profound?
Almost certainly not at first. Start with the spreadsheet method, graduate to a 29 to 99 dollar tool when manual running gets tedious, and consider enterprise platforms only if you manage multiple firms or markets. The method matters more than the software.
Do You Know What AI Says About Your Firm?
We run AI visibility tracking for personal injury firms across Los Angeles, and fix what the data reveals.
AI Visibility For PI Law Firms
Rank tracking taught a generation of firms that what gets measured gets managed. AI visibility is the same discipline with a new scoreboard: frozen questions, scheduled runs, dated logs, and a short list of metrics that tie mentions to fixes.
Start tonight with fifteen questions and a spreadsheet. In three months you will have something almost no firm in your market possesses: proof of where you stand, evidence of what works, and a map of exactly who you need to displace. The machines are already answering questions about your market. The only choice is whether you are watching.
Sources and References
- Zapier: the best AI visibility tools, pricing and coverage compared
- Surfer: AI visibility tools compared, including API versus real interface data
- ESEO Space: AI visibility tools for agency delivery and the SIGIR consistency research
- SEO Noble: AI search optimization for personal injury lawyers
- SEO Noble: personal injury lawyer SEO in Los Angeles
