South Florida Personal Injury AI Visibility Index

Many marketers are adding schema markup and llms.txt files to get their firms into AI answers. In South Florida personal injury, we found Organization schema less often on the firms AI recommends most than on the firms it ignores. What separated them was what the rest of the web says about them.

Key findings

One firm leads everywhere. After that, the platforms disagree.

Morgan & Morgan is first or second on every platform, and Panter, Panter & Sampedro is in the top five on three of them and the top ten on all four. Below them, each assistant builds a different shortlist. Podhurst Orseck is second on ChatGPT but absent from Gemini; Colson Hicks Eidson — second in our ChatGPT-only first edition — is named by just two platforms. A firm that checks only ChatGPT sees a different market than its clients do.

Consensus ranking — top 10

Average AI Visibility score across the four platforms (a firm not named on a platform counts as 0 there). Scores run from 0 to 100.

Full ranking of all 119 named firms, with each firm’s type and per-platform scores: download the CSV.

Full dataset, free to reuse under CC BY 4.0: all 720 AI answers with the sources each one cited, the 60 questions, firm scores and cited domains, on GitHub and Hugging Face. The answers are raw AI output and have not been fact-checked.

Not every firm AI names fits the question

We checked every one of the 119 names the assistants gave. 81 are personal injury firms with a South Florida office (9 of them national firms with local offices). The other 38 — nearly a third — do not fit the question that was asked: 22 have no South Florida office at all (mostly national mass-tort firms based in New York, Chicago or elsewhere), and 16 are South Florida firms whose main practice is something else, such as large commercial firms, insurance-defense firms or immigration practices. Every firm in the consensus top 25 is a personal injury firm with a South Florida office.

AI reads the directories, not the billboards

When the assistants searched the web before answering, they leaned on legal directories. Every Perplexity answer that cited sources cited at least one directory; most Claude and ChatGPT answers did too. For a firm, that means the profile it keeps on Super Lawyers, Avvo, Lawyers.com or FindLaw can matter more to an AI answer than its own advertising — and a firm missing from those sources is missing from the material the AI reads.

Gemini returns its sources as Google redirect links, so its directory share could not be measured in this edition.

Google Maps visibility does not carry over

We started from the market as a client sees it on Google: the top 100 Google Maps results for “personal injury lawyer” in each of the four cities. Firms with many Google reviews are not reliably named by AI. Four of the ten most-reviewed firms on Google Maps were named by none of the four assistants. In the other direction, five of the ten firms the assistants name most often do not appear in Google Maps’ top 100 in any of the four cities.

What the top 10 do differently

We compared the ten firms AI recommended most in an early cut of the data (nine are in the final top ten; Brian D. Guralnick Injury Lawyers, now 13th, was included instead of Goldberg & Rosen) with ten South Florida firms that the assistants named at most once (nine never, one in a single Perplexity answer) — firms with strong Google profiles (roughly 400 to 1,300 reviews each), spread across the same four cities.

The pattern is consistent: what separates the firms AI recommends is not their website’s technical polish or their Google review count — it is their presence beyond their own site. They appear on more legal directories, especially peer-review and award-based ones, and they carry more authority on the wider web. One top-five firm’s website even blocks automated access. When the assistants named a top-ten firm, a legal directory was among the cited sources in at least two thirds of those answers.

Ten firms against ten: this shows association, not cause. Directory presence was checked by search and is approximate; it shows whether a firm appears on a directory, not how complete its profile is.

What about schema and llms.txt?

Two technical fixes are often sold as the way into AI answers. Our data does not support either as the deciding factor in this market.

• Organization schema: we found it on only half of the AI top 10 (one of those sites blocks automated access, so its markup could not be checked), against nine of the ten rarely-named firms. Structured data is good housekeeping for search engines; it did not separate the firms AI recommends.
• llms.txt: seven of the top 10 had one, against five of the comparison group. Google says its AI features do not use llms.txt, so the likelier reading is that firms with active marketing teams tend to have one — not that the file earns the recommendation.

Both are associations in ten-versus-ten samples, reported because they are often asked about — not as evidence that either file helps or hurts.

What changed since Edition 1

Edition 1 (May 2026) asked ChatGPT 40 questions once each and ranked every firm ChatGPT named. Edition 2 repeats those 40 questions word for word, adds 20 city-specific questions, asks each question three times, and covers four platforms instead of one.

• Morgan & Morgan remains first on ChatGPT.
• Podhurst Orseck stays near the top on ChatGPT (third in May, second in October); Colson Hicks Eidson moves from second to fifth.
• New on ChatGPT’s top five: Rosen Injury Law. New among the cross-platform leaders: Lytal Reiter and Kelley | Uustal, neither named in Edition 1.

Correction and method note. Edition 1 was run on ChatGPT through OpenAI’s API (gpt-5-mini with web search), not ChatGPT-4o as its report stated. Its “116 firms” were the names ChatGPT mentioned, including three legal directories and several firms outside South Florida — not a list of South Florida firms. Its scores used an earlier formula and are not directly comparable with Edition 2 scores; we compare positions, not points.

Methodology

• Firm list: the top 100 Google Maps results for “personal injury lawyer” in Miami, Fort Lauderdale, Boca Raton and West Palm Beach (October 2026); law-firm listings with at least 10 Google reviews, with offices of the same firm merged by website — 283 firms.
• Questions: 60 — the 40 Edition 1 questions verbatim (direct, problem-based, comparative, authority and scenario questions) plus 20 city-specific questions, five per city.
• Platforms: ChatGPT (gpt-5-mini), Gemini (3.1 Flash Lite), Claude (Sonnet 5.5) and Perplexity (Sonar), each through its API with web search on and no prior conversation. Each question was asked three times on each platform — 720 answers. API answers can differ from the consumer apps.
• Scoring: the AI Visibility score (0–100) combines how often a firm is named (35%), how often it appears in the top three (30%) and how substantively it is described when named (35%, weighted by how often it is named). Firm names were matched across spellings, offices and former names, and every named firm was checked for a South Florida office and practice area.
• Limitations: one data-collection window; API rather than app answers; Gemini sources not attributable. Being cited on a directory is associated with being recommended; this study does not prove that it causes it.

What firms can do

Check whether AI names your firm for the questions your clients ask, on every platform — not just one. Make sure your profiles on the directories AI reads are complete and accurate, and that your own site clearly answers the questions clients ask. Earn mentions on the sites AI already trusts. Avoid shortcuts: paid placement schemes and fake reviews put a firm at risk and are not what AI rewards.

To see where your firm stands, request a free AI visibility snapshot below.