SeeWAIS
When families in Cleveland need a divorce or custody attorney, does AI name your firm — or a competitor's?
We put the same 25 blind buyer questions to ChatGPT, Gemini and Claude about a real family law firm in Cleveland, OH — no brand name in any of them — then fact-checked every claim the assistants made against the business's own site. Below is what came back: every number from that measurement, nothing modelled.
recommended in 42 shortlist searches
of 233 candidate businesses
visibility score
wrong factual claims from AI
The summary
#12 of 233 candidate businesses, with 4 wrong factual claims — two of them attorneys the site does not list. Fee questions came back 0 of 9, and individual clients were named in 2.0% of their questions against 13.3% for business ones. A rank is not visibility: answers are built line by line, and the consumer side is barely in them.
attorney fees — 9 questions about what a lawyer costs, not one of them naming the firm. AI still quoted contingency rates from 33 1/3% to 40% that the site never publishes.
Three of the real buyer questions we asked
Blind questions, quoted verbatim from the run — the kind a real customer types, with no brand name in them.
- “What are the top-rated law firms in Cleveland for family law?”
- “What should I look for in a Cleveland commercial real estate lawyer to handle a lease negotiation?”
- “What are the attorney fees for a Cleveland firm to handle an uncontested divorce?”
One AI answer we fact-checked
“The firm has 30+ attorneys”
ChatGPT (searching the web) · brand-audit answer N02
What the fact-check found: Website states 'With 30+ attorneys' but the attorney directory lists only 22 attorneys.
Where AI names this firm — line by line
| Revenue line | Named / asked | Mention rate |
|---|---|---|
| attorney fees | 0 / 9 | 0.0% |
| legal services for businesses | 2 / 15 | 13.3% |
| legal services for individuals | 1 / 51 | 2.0% |
What this means for a firm owner
Fee questions were probed 9 times and the firm was named in none of them. AI still quoted figures — 33 1/3% for pre-lawsuit settlements, up to 40% once litigation starts — with nothing on the site to back them. The report sequences its work on this line first: that is where revenue is being lost.
Recommended in 3 of 42 shortlist searches (7.1%): 1 of 25 questions on ChatGPT, 2 of 25 on Gemini, none on Claude. The next moves are machine-readable service and price pages, plus the profiles the answers are built from.
Legal services for individuals: 1 of 51 questions (2.0%); legal services for businesses: 2 of 15 (13.3%) — named more than six times as often. AI describes the business practice and skips the consumer one: the individual-client pages need their own structured presence instead of sharing the firm-level page.
AI built its answers about this market from 111 third-party sites and the firm appears on 1 of them. The fact-check found 4 wrong claims, including two attorneys the site does not list. The profiles AI already reads are free to claim and take minutes — the report sequences that work first.
What a report for your firm looks like
The same measurement, run on your domain: every AI answer quoted in full, every factual claim checked against your site, the third-party sources the answers are built from, and a ranked action list with ready-to-paste code. The full method is published; the complete report behind every number on this page is public.
Measured September 2026. The measured business is pseudonymised; every number is real and traceable to the published run.