AI Visibility Audit: How a Law Firm Went From 11% to 40% in 27 Days

What is AI visibility?

AI visibility is how often AI tools mention your business when someone asks a question you should be the answer to. If a person types "I need a lawyer after a car dealership cheated me" into ChatGPT, does the answer name your firm? That's the question.

It matters because more buyers now start with an AI tool instead of a search engine. They ask a full question, get a short answer, and often act on the names in it. If you aren't named, you aren't in the running.

Google rankings still matter. AI visibility is a second place where you can be found, and it works a little differently. Google shows a list of links. An AI tool gives one answer with a few names, and the buyer rarely scrolls for more.

What does an AI visibility audit measure?

An audit asks a set of questions across several AI models and records what comes back. A good one tracks five things:

• Brand visibility. The share of answers that mention your business at all. If you ask 100 question and model combinations and you appear in 11, your visibility is 11%.
• Sentiment. Whether the mention is positive, neutral or negative. Being named is good. Being named as the one to trust is better.
• Which models name you. ChatGPT, Gemini, Perplexity, Claude, Copilot and others use different sources, so you can be strong in one and missing in another.
• Share of voice. How often you appear compared with competitors. This is your rank among the businesses the tools name.
• Whether your website is cited. Some tools show sources. A link to your site is worth more than a name alone, because it can send a visit.

Each of these tells you something different. High visibility with weak sentiment means people hear about you but not for good reasons. High visibility in one model and none in another points to a source problem. Being named without your site cited means a competitor's or a directory's page may be getting the traffic.

What did a real AI visibility audit show?

We ran two audits for a consumer law firm on the keyword "auto fraud," about four weeks apart: May 14, 2026 and June 10, 2026. That's 27 days. Each audit checked 10 AI models. Here is how the numbers moved:

On June 10, Perplexity's own output said the firm was "the primary authority in the provided search results, appearing in website listings, practice area descriptions, and video content specifically targeting auto fraud keywords."

That's one firm, one keyword and one 27-day window. We can't say that every business will see the same jump, and an audit alone doesn't prove what caused each change. What it does give you is a clear before and after, so you can see whether your work is moving the needle. Without a baseline, you only have a feeling.

There's a second data point worth noting. In a separate test, we typed, "I live in Arizona and a car dealership screwed me, I need a lawyer," into ChatGPT. It named the firm's Tucson office as an option. That's the kind of everyday question your buyers actually type, and it's why we test with full sentences, not only short keywords.

How can you run a basic AI visibility audit yourself?

You can do a useful first version in an afternoon with no special software. Here is how:

1. Write 10 to 20 questions your buyers ask. Use real language. Not "personal injury attorney" but "I was hit by a driver who ran a red light, what should I do?" Include questions with your city, questions without it, and questions that compare options.
2. Add a few brand questions. "Is [your company] any good?" and "What does [your company] do?" show how AI tools describe you.
3. Pick the tools. Start with ChatGPT, Gemini, Perplexity, Claude and Microsoft Copilot. Add others your buyers might use.
4. Ask each question in each tool. Use a fresh chat each time. If you can, log out or use a private window so your own history doesn't skew the answer.
5. Record the result in a spreadsheet. For each answer, note whether you were named, how you were described, who else was named, and whether your website was cited.
6. Repeat each question a few times. AI answers change from one run to the next. One answer is an anecdote. Several runs show a pattern.
7. Calculate your numbers. Visibility is the share of answers that name you. Share of voice is how you compare with the other names.
8. Repeat monthly. One audit is a snapshot. Two or three is a trend.

Two cautions. First, the answers depend on the exact wording, the location and the tool's current settings, so treat any single result as a sample. Second, Google queries aren't the same as AI prompts. A phrase that ranks well in Google may not get you named in ChatGPT, and the reverse is true too. Test both.

If you'd rather not do this by hand, Rob built a scanning tool called AI Search Scan that runs these checks across AI models. It's one option, not the only one, and the manual method above works fine to start. Either way, what counts is that you measure the same questions the same way each time.

Why isn't visibility the goal?

Because being named doesn't pay the bills. A mention that never leads to a visit, a call or a booking is a vanity number.

Think of visibility as the top of a funnel:

1. The AI tool names you. (Visibility)
2. The reader clicks to your site or searches your name. (The click)
3. Your page answers their question and makes the next step easy. (The page)
4. They call, book or fill out the form. (The call)

Each step can leak. You can be named and not clicked. You can be clicked and land on a page that doesn't match the question. You can have a great page and a phone nobody answers. An audit shows you step one. The rest takes tracking and good pages. Our guide to how to get traffic from ChatGPT covers steps two and three.

The real-world numbers show the link between the two. From May 6, 2025 to September 3, 2026, we tracked 2,476 AI-referred visits to 500 identified pages across 18 websites. They came from ChatGPT, Perplexity, Gemini, Claude, Copilot and others. ChatGPT sent 87% of the visits that could be traced to a source. One consumer law firm site alone received 1,126 AI visits across 123 pages. Its homepage got 544, and its auto fraud attorney page got 68. That page also ranks in Google for "auto fraud attorney" and "best lawyer to sue car dealership." A page on what happens when a dealer can't fix your car under warranty ranks around position 3 for that exact question.

The pattern is that the same pages that answer real questions well tend to show up in both places. You can read the full numbers in our AI referral traffic report.

How does AI visibility tracking tie to Google Analytics?

Visibility tracking tells you what AI tools say. Google Analytics tells you what people do after they read it. You need both.

In Google Analytics 4, you can look at your traffic sources and find visits that came from AI tools. Look for referrers such as chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai. Then:

• Group them together. Create a custom channel or segment called "AI search" so you can watch the trend in one line.
• See which pages they land on. These are the pages AI tools trust enough to send people to. Improve them first.
• Connect them to conversions. Mark form submissions, calls and bookings as key events so you can see which AI visits turn into leads.
• Remember what you can't see. Some AI visits arrive with no referrer and show up as direct traffic. In our data, ChatGPT sent 87% of the visits that could be traced to a source, which means some visits could not be traced at all. Treat your AI number as a floor, not a ceiling.

Put the two views side by side each month. If visibility goes up and visits don't, look at whether your site is being cited and whether your pages match the questions. If visits go up and calls don't, look at your page and your phone follow-up.

What helps your visibility improve?

No one controls what an AI tool says, and anyone who promises a placement is overselling. But the things that tend to help are in your hands:

• Pages that answer questions directly. Put the answer in the first few lines, then add detail.
• Consistent business details. Your name, address, phone number and descriptions should match on your website, your Google Business Profile and your other listings.
• Real reviews and mentions. AI tools learn from what others say about you, not only what you say about yourself.
• Video and other formats. In the audit above, Perplexity pointed to video content as one reason it named the firm.
• Clear author and company information. Pages with real people behind them are easier to trust.

Our page on AI search optimization explains these in more detail. If you want help, look at our AI search service. To start with a smaller step, read how to get your business on ChatGPT.

How often should you run an audit?

Monthly is a good rhythm for most businesses. AI answers shift as the tools update and as new pages appear, so a single audit goes stale fast. The law firm audits above were 27 days apart, which is about right for seeing movement without drowning in data.

Keep three things the same each time: the questions, the tools and the way you score the answers. If you change the questions every month, you can't tell whether a change in your numbers came from your work or from your new wording. Add new questions as separate rows, and keep your original set as the baseline.

Also keep a short log of what you changed on your site between audits: a new page, an updated profile, a batch of new reviews. When the numbers move, that log helps you guess why. It won't prove cause, but it beats guessing from memory.

What mistakes should you avoid with AI visibility tracking?

1. Testing only your brand name. Buyers who already know you aren't the issue. Test the questions strangers ask.
2. Trusting one answer. Run each question several times and look for the pattern.
3. Treating Google queries as AI prompts. Short keywords and full questions behave differently. Test full sentences.
4. Counting mentions and ignoring sentiment. Being named in a bad light doesn't help.
5. Stopping at visibility. If you aren't tracking visits and leads, you can't tell whether this is worth the time.
6. Expecting a guarantee. AI answers shift. You can improve your odds, but you can't promise a spot.

Where should you start?

Run the basic audit above this week. It takes an afternoon and it tells you where you stand today. Then open Google Analytics and see how many visits already come from AI tools, and which pages they land on. Those two views, visibility and visits, are your baseline.

If you want a plain-English read of where your business is leaking leads, from search and AI answers to follow-up, our Growth Leak Audit is the next step.