Legal practice management software, measured across four AI engines

Thirty-nine answers, four engines, one legal practice management question set. Seven vendors appeared regularly and the spread between them is narrower than in most categories we have measured.

Clio took 31 of 39. Six other vendors were named in a third or more of all answers.

This is an unusually crowded answer surface. In our own agency category five firms shared 19 namings across 120 open questions. Here seven vendors share more than 130 namings across 39 answers, which means most answers listed three or four names rather than one. A crowded list is easier to enter and harder to lead.

The most useful single finding in this sweep is not a ranking. ChatGPT named a product identity that the anonymised vendor retired in 2023 five times, and named the vendor's current brand zero times.

The engine was answering from material written before the rename.

A buyer reading that answer is being pointed at a product they cannot buy, under a name the vendor has spent three years retiring. Every one of those five answers is a deal the vendor did not enter under its own name, and a conversation the sales team never saw start.

The mechanism is mundane. Engines retrieve what is readable, and the internet keeps an old name for far longer than the company does. Reviews, forum threads, comparison articles and directory listings written before a rename stay indexed and stay retrievable. Unless the vendor publishes material that explicitly ties the old identity to the new one, the engine has nothing telling it to update.

This is one of the few problems in AI search visibility with a direct remedy, because the engine is not making a judgment about the vendor. It is repeating the most readable thing it found.

The material that changes it states both names in the same sentence, on pages an engine has reason to open: product documentation, a migration or naming note, comparison pages that use the current brand, and answers to the specific operational questions buyers ask. The point is not a press release. It is that a retrieval system encountering the old name finds the new one attached to it.

We publish the method we use for measuring this at github.com/Broadcastwell/audit-method so a vendor can run it without hiring anyone, and our own score against it, which was 0 of 40, at our own visibility page.

In a category where most answers name three or four vendors, the question changes. It is no longer whether you can enter the list, it is whether you enter it under the right name and against the right criteria.

Being named three times out of 39 in a list this crowded is a different problem from being named zero times in a concentrated one. The material exists, buyers are asking, competitors are being described in detail, and the vendor is absent from the description rather than absent from the market. That gap closes with publishing that answers the question rather than with more brand spend.