The RecordNo. 04

AI does not rank your firm, it answers questions

C

By the CW Team

September 2026. 4 minute read.

Someone at your firm typed "best estate planning attorney in Charlotte" into ChatGPT, your name came up, and the room felt good about it for a week. Then a client mentioned she had asked something close to that and got three other firms, none of them yours. Both things happened. Neither one was a fluke.

This is the part of AI visibility that catches firm owners off guard. There is no ranking table sitting behind the answer. There is no position three that you hold on Tuesday and still hold on Friday. Each answer is built in the moment, from whatever the model retrieved for that exact phrasing, and a small change in phrasing pulls a different set of sources.

The same firm, two answers

Ask about "best estate planning attorney in Charlotte" and the model leans on directory pages, rankings, roundup articles, the sources that treat law firms as a list to be sorted.

Ask "my mother died without a will in North Carolina and my brother is on the deed, do I need a lawyer" and the model is somewhere else entirely. It is reading explanatory pages about intestate succession, forum threads, state bar material, firm blog posts written by people who handle that exact mess. Whoever explained the problem well is in the answer. Whoever bought their way onto a best of list is not.

Same city. Same practice area. Same client, honestly, one week apart in her own thinking. Two different sets of firms named.

What clients actually type

The general question is the one you test, because it is the one you would ask about your own firm. It is not the one that brings you work.

Real questions are longer and messier and they carry the fear in them. "I got fired two days after I filed a workers comp claim." "How much does it cost to fight a first DUI in Ohio." "The insurance company offered me eleven thousand dollars, is that low." "Can my ex move out of state with our kids if we have joint custody." "My contractor stopped showing up and he has forty percent of the money."

Those are the questions where somebody is close to hiring. They are also the questions where the answer depends less on directories and more on whether anyone from your firm has ever written plainly about that situation. A person who asks about a low settlement offer is further along than a person browsing for the best injury lawyer in town. If your name appears for the browsing question and vanishes for the settlement question, you are visible in the least valuable place.

One check is an anecdote

This is why checking your firm once tells you almost nothing. You tested one phrasing on one day in one assistant. The result is real and it is also unrepresentative.

Visibility only becomes a fact when you look across a wide set of questions at once, the way clients actually spread across them. That is why Casewell measures against more than 500 questions rather than a handful. What comes out is one number, the share of those recorded answers that name your firm. The number matters less than what sits underneath it, because the same score can be built from strength across the board or from strength in one corner and silence everywhere else. Split by the kind of question being asked, those two firms look nothing alike. From the inside the second one still feels like winning, because every check you personally run happens to be inside the corner you already own.

The gaps are the useful part

When you look at a whole set of questions, the pattern that matters is not where you appear. It is where you do not, next to where you do.

A firm that shows up for every general question in its city and disappears on procedural questions has a content problem, not a reputation problem. A firm that shows up on procedural questions and vanishes on the comparison ones, the "who is the best" and "how do I choose" type, has a third party problem. It is not present in the places that get read when someone wants a list. Those two gaps take completely different work to close, and you cannot tell them apart from a single check.

The gaps also tell you which practice areas are quietly unrepresented. Most firms find at least one area they consider core where AI does not associate them with the work at all, because everything published about it lives on an internal page nobody links to and nothing outside the firm confirms it.

What this means for your firm

Stop asking whether you are visible in AI. The question has no answer, because visibility is not a state your firm is in. Ask which questions name you, which ones do not, and whether the ones that do are anywhere near the moment a person decides to call a lawyer. That list is short enough to read in a meeting and specific enough to act on, which is more than most marketing reports manage.

See what the assistants say about your firm.

Or read what a Casewell engagement includes and what it costs.

All entries in The Record