How to Audit Your Law Firm's AI Visibility Yourself in One Hour
Twenty questions, four engines, one spreadsheet. Run each question in a fresh session, log every firm named and every source linked, and you have a citation rate and a list of who is cited instead of you. Here is the exact procedure we use, scaled down so a firm can run it before ever calling an agency.
Short answer: Write twenty questions a client would ask about your practice area in your city, run each one in ChatGPT, Perplexity, Gemini and Google (for the AI Overview), each in a fresh session, and log every firm named and every source linked. Your citation rate is the share of the eighty answers that name you. The firms that appear instead of you, and the sources behind them, are the audit. It takes about an hour and it tells you more than any ranking report.
Why run it yourself first?
Because the result is not intuitive. Firms that rank on page one routinely discover they are absent from every AI answer, and firms with modest rankings discover Perplexity names them for procedural questions. Only 38% of AI Overview citations come from the organic top ten (Soar), and ChatGPT names brands in under 1% of answers, so most firms start at zero and do not know it. An hour with a spreadsheet replaces the guess with a number.
Step 1: write the twenty questions (15 minutes)
Cover the client's decision path, not your service list. Five groups, four questions each, with the city attached where a client would attach it.
| Group | Example for a Pasadena estate firm |
|---|---|
| Do I need a lawyer | Do I need a lawyer to make a living trust in California? |
| Process and timeline | How long does probate take in Los Angeles County? |
| Cost | How much does a living trust cost in Pasadena? |
| Courthouse and local | What happens at a probate hearing at the Pasadena courthouse? |
| Who should I hire | Who is a good estate planning attorney in Pasadena? |
Write them the way a person types, not the way a lawyer would. Include one or two in the language your clients use if it is not English.
Step 2: run them (35 minutes)
For each question, in each engine, in a new chat or incognito session so earlier answers do not shape later ones:
- ChatGPT with search enabled. Note whether it cites at all; often it will not.
- Perplexity. Note the numbered sources.
- Gemini. Note the sources panel.
- Google. Note whether an AI Overview appears and which links it carries, then note the map pack and the top organic results separately.
Log four things per answer: the firms named, the sources linked, whether you appear, and the position if you do. A spreadsheet with one row per question per engine is enough.
Step 3: read the result (10 minutes)
- Citation rate. Rows where you are named divided by eighty. Most firms start between 0% and 10%.
- Who is cited instead. Usually two or three local firms and several directories. Those firms are your actual competitors for AI answers, whatever the rankings say.
- Which sources carry the citations. Avvo, Justia, Yelp and the court's own pages appear constantly. If a competitor's own site appears, open the cited page and note where the answer sits on it.
- Which question groups you are absent from. Absence on courthouse and cost questions is the most common and the fastest to fix, because those pages usually do not exist yet.
Step 4: check your own signals (the remaining minutes)
Five checks, all free:
- Entity. Open your State Bar profile, Google Business Profile, Avvo and LinkedIn side by side. Does the firm name, address, phone and practice list match character for character? See the directory consistency checklist.
- Schema. Paste a practice page URL into a schema validator. Is there a LegalService block, a Person block, FAQPage on question content?
- Crawler access. Read
yourfirm.com/robots.txt. Is anything underUser-agent: *disallowed that should not be, and is there a blanket AI block? See which AI crawlers to allow. - Answer position. Open your main practice page and find the first sentence that answers a client question. If it is below the fold, that is the rewrite.
- Reviews. Count reviews on Google and Avvo from the last ninety days. Fewer than six is below the velocity that moves citations.
What the one-hour version misses
Randomness, breadth and repetition. AI answers vary run to run, so a single pass overstates or understates a firm; forty to sixty questions run twice a week apart is the minimum for a stable number. And the signal audit of the cited competitors takes longer than an hour when done properly. That fuller version is the AI citation audit we run for free for Los Angeles firms, with a written report. The one-hour version is enough to know whether you need it.
What to do with the result
If your citation rate is near zero and the entity checks fail, the fixes are in the AI search visibility guide: align the listings, add the schema, allow the crawlers, rewrite the practice page answer-first, and start the review cadence. If the entity checks pass and you are still absent, the gap is content: the courthouse page, the cost page and the client-question pages that do not exist yet. Either way, keep the spreadsheet; rerunning it in ninety days is how you will know the work moved anything.
Check your firm
See whether ChatGPT, Perplexity, Gemini and Google AI Overviews name your firm for your practice area and city.
Free AI visibility check