Free AI visibility audit for law firms
This free AI visibility audit writes eight questions a client in your city might ask an AI assistant about your practice area. You ask ChatGPT, Gemini and Perplexity, log whether each one named your firm, and get a mention rate, a first-mention rate and a per-assistant breakdown. It runs in your browser, and nothing is sent anywhere.
Tell the tool about your firm and it writes the questions. You ask them, log what each assistant said, and get your result. Plan on about 20 minutes for all 24 answers.
Run the AI visibility audit
1. Ask each assistant these questions
Use a private or logged-out window so your own history doesn’t sway the answer. Copy each question into ChatGPT, Gemini and Perplexity.
2. Log what each assistant said
Pick “Named first” if your firm was the first one recommended, “Named” if it appeared anywhere, and “Not run” for anything you skipped. You need at least 3 answers from one assistant for a score.
| Question | ChatGPT | Gemini | Perplexity |
|---|
0%
By assistant
What to do next
Want to know why you’re missing?
Send your result and William will reply personally with what he’d look at first. Your numbers go with the message.
How the audit works
It measures one thing: how often an AI assistant names your firm when a client in your city asks for a lawyer like you.
The method, step by step
- Eight questions. They cover the ways clients tend to ask: “best”, “recommend”, a real situation, reviews, top-rated, affordable, free consultation and experience. Each one names your practice area and city.
- Three assistants. ChatGPT, Gemini and Perplexity, each asked in a private or logged-out window so your own history doesn’t sway the answer.
- Your log. For each answer, you mark whether your firm was named first, named anywhere or not named. Anything you skip is left out of the maths.
- The result. Your mention rate is the answers that named you divided by the answers you logged. Your first-mention rate is the same, counting only answers where you were recommended first. Each assistant gets its own rate once it has at least three answers.
A worked example
This example uses made-up numbers for a sample firm, to show how the maths works.
24 answers, logged and scored
A sample estate planning firm runs all eight questions in all three assistants, so it logs 24 answers.
| Assistant | Named | Named first | Answers logged |
|---|---|---|---|
| ChatGPT | 6 | 3 | 8 |
| Gemini | 3 | 1 | 8 |
| Perplexity | 5 | 1 | 8 |
| Total | 14 | 5 | 24 |
Its mention rate is 14 of 24, which the tool shows as 58. Its first-mention rate is 5 of 24, shown as 21. Gemini names it least, in 3 of 8 answers, so the tool suggests looking at which firms Gemini does name and what their sites and profiles say.
What this audit can’t tell you
It’s a snapshot of what three assistants told you on one day about eight questions.
The limits, in full
- Answers change. The same question can get a different answer tomorrow, or in another city. Re-run it monthly and compare.
- It isn’t every question. Clients ask in many ways. Eight questions give a fair sample, not a complete picture.
- It doesn’t say why. It shows whether you’re named, not what the assistant read. The suggestions point to the usual causes, and our guide to AI recommendations for law firms covers each one.
- It doesn’t check your site. For the technical side, such as whether your robots.txt lets AI crawlers in, use our law firm website audit.
- Google’s own view. Google says there’s nothing special to add for its AI features beyond normal good SEO practice, in its guidance on AI features and your website.
Who built this, and a disclosure
William Schneider built this audit for XYR Partners. He also built CompeteScan, an AI visibility tracker for ChatGPT, Gemini and Perplexity that runs this kind of check automatically. You don’t need CompeteScan to use this free tool, and nothing here sends data to it.
Frequently asked questions
Why do I have to run the questions myself?
Because AI answers depend on who's asking, where and when, and the assistants don't offer a free, fair way to check them automatically. Running them yourself in a private window shows you what a new client in your city would most likely see.
How many questions should I run?
All eight in each assistant if you can, which is 24 answers. You need at least three answers from one assistant to get a result, but more answers give a steadier picture.
Does this tool send my firm's name anywhere?
No. The questions are written in your browser and your log stays there. Nothing is sent unless you choose to email your result to William using the form.
My firm wasn't named at all. Is something broken?
Usually not. AI assistants tend to name firms they can find clear, consistent information about across your site, your Google Business Profile and other sources. Start with the fixes the tool suggests and re-run it in a month.