We asked 3 AIs for a personal injury lawyer in Austin. 57 firms came up.

On September 27, 2026, we put ten questions about finding a personal injury lawyer in Austin to ChatGPT, Perplexity and Google's AI Overviews. We requested 30 results and received 27 answers. Together, those answers named 57 distinct firms. Seven firms appeared across all three services somewhere in the test.

The interactive Austin report lets you inspect the questions, named firms and cited sources. The variation is easier to see in those answers than in a single visibility score.

Ten questions about one market

We used questions about a practice area and city, with variations a prospective client could ask. They covered general personal injury recommendations, car and truck accidents, motorcycle accidents, slip-and-fall cases, wrongful death, affordability, reviews and Spanish-language service.

These were test prompts, not a ranked list of search demand. A broad query such as “best personal injury lawyer in Austin TX” tests a different request from “Spanish speaking personal injury lawyer in Austin.” The latter adds a specific requirement. A firm that fits that requirement has a reason to appear even if it is absent from a broad answer.

Every question went to all three services. Three requested results did not produce returned answers, so the denominator for the overall mention counts is 27. A missing result cannot tell us which firms an AI would have named.

The most-mentioned firm was absent from one service

FVF Law Firm appeared in 11 of the returned answers: eight from ChatGPT and three from Perplexity. It did not appear in the Google AI answers captured in this test. Sandoval & James appeared eight times, spread across all three services.

Firm	ChatGPT mentions	Perplexity mentions	Google AI mentions	Total
FVF Law Firm	8	3	0	11
Sandoval & James	3	1	4	8
Herrman & Herrman	2	1	3	6
Byrd Davis Alden & Henrichson	4	1	1	6

These counts come from the September 27 sample. They measure appearances in the captured answers, not market share, case volume or the quality of a firm's work. Eleven appearances also do not mean eleven people saw or acted on a recommendation.

For a marketing team, the FVF result is a reason to keep separate columns for each service. An aggregate score would hide its absence from the Google AI results. Our city-by-city checking guide explains how to make that comparison for your own practice.

Both directories and firm websites appeared as sources

The report lists Super Lawyers with nine source appearances, Avvo with eight, Expertise.com with six and Justia with five. Individual law-firm websites also appear among the cited sources.

That gives a firm somewhere concrete to investigate. Open the cited page and read what it says about the firms in the answer. Check whether your own listing exists, whether its details are accurate, and whether the page's inclusion criteria apply to you. A cited directory is a research lead; it is not a guaranteed route into the next answer.

The report also shows different source mixes by service. It would be a mistake to turn this sample into a rule that every AI prefers directories, or that a firm's own website no longer matters. Citations identify sources attached to answers. They do not expose the complete process that selected each name.

What this test can tell a firm

A single-city snapshot gives you examples to inspect and a method to repeat. It cannot establish how recommendations change over time, what caused a firm to appear, or how many prospective clients use these answers.

The sample was collected through Olostep on one date. Treat it as that captured service output, not a promise that every person opening a consumer app would have seen identical wording. The comparison holds the questions constant; it does not control every setting or expose every model decision.

If your firm serves Austin, start with the questions that match your practice and compare the answers directly. If you serve another city, generate a local question set. Keep the wording, date, returned answers and failures so a later run has a usable baseline.

The law-firm recipe includes the sample and offers a free beta check of up to five questions across the three services. Use the results to choose a specific profile or page to improve, then follow a consistent review schedule.

Questions

How many answers did the Austin test collect?

Ten questions went to three AI services on September 27, 2026, for 30 requested results. Twenty-seven answers returned. Those answers named 57 distinct firms.

Did all three AI services recommend the same firms?

Seven firms appeared across all three services somewhere in the test. That does not mean they appeared together for every question, or that the services agreed on order.

Does an AI mention show that a law firm is better?

No. This test measures which names appeared in a set of AI answers. It does not evaluate legal work, suitability for a case, or whether anyone contacted a firm.