Twenty-three of the fifty largest personal injury firms in the country are never mentioned, not once across ten buyer-intent questions, on at least one major AI platform. Fifteen are missing from two or more. These are firms with national ad campaigns and market-defining verdicts, absent from the channel where a growing share of their future clients now begin.
| Platform | Visibility | Share visible |
|---|---|---|
| Gemini | 92% | |
| ChatGPT | 80% | |
| Claude | 80% | |
| Perplexity | 72% | |
| Grok | 72% | |
| Google AI Overviews | 24% |
| Named on all five chat platforms | 27 firms | |
| Missing from exactly one | 8 firms | |
| Missing from two or more | 15 firms |
When we asked each platform directly about each firm by name, using the prompt “is [firm] a good law firm?”, the AI knew essentially every single one: histories, practice areas, notable verdicts. Recognition across the five chat platforms was 98 to 100%. Then we compared that against the buyer-intent questions, where a potential client asks who to hire. Across 51 firm-platform pairs, the same AI that can recite a firm's biography never once offered that firm's name to a prospective client. Twenty-three firms sit in this gap on at least one platform.
| Recognized when asked directly | 100% | |
| Recommended on every platform | 54% |
If AI platforms all drew from the same picture of the world, a firm visible on one would be visible on all. That is not what the data shows. Twenty percentage points separate the most generous platform (Gemini, 92%) from the stingiest (Perplexity and Grok, 72%). Firms routinely dominate one engine while being absent from another.
Even the friendliest platform misses 4 of the 50 biggest firms in the country entirely.
Ten top-50 firms never appear in ChatGPT answers, including nationally famous names in their home markets.
Both lean on live retrieval and citations. 28% of top firms are invisible on each.
Several firms that appear everywhere show only one or two mentions on their weakest platform.
We graded each firm by how many of the six platforms mention it at least once. The distribution says the market is wide open: barely half of the biggest firms in the country hold strong visibility across the board, and the rest are exposed. One firm, despite ranking among the largest injury practices in America, never appeared in a single answer on any platform tested.
| Tier A, visible on 5 to 6 platforms | 28 firms | |
| Tier B, visible on 3 to 4 platforms | 13 firms | |
| Tier C, visible on 1 to 2 platforms | 8 firms | |
| Tier D, visible nowhere | 1 firm |
Each of the six AI platforms retrieves, verifies, and cites differently, and each produces a different map of the legal market.
| Platform | Visibility | Share | Retrieval |
|---|---|---|---|
| ChatGPT gpt-4o | 80% | Retrieval plus trained knowledge | |
| Claude claude-sonnet-4-6 | 80% | Trained knowledge plus retrieval | |
| Gemini gemini-3.5-flash | 92% | Google index plus retrieval | |
| Perplexity sonar | 72% | Live retrieval plus citations | |
| Grok grok-4.3 | 72% | Live retrieval plus citations | |
| Google AI Overviews google-ai-overview | 24% | Google SERP feature |
The same test behind this study can be run for your firm: your market, your practice areas, your competitors, across all six platforms.
Get your firm's AI visibility scorecardIn July 2026, Faneros tested the 50 largest US personal injury firms, selected from public rankings, against 12 questions per firm: 10 unbranded buyer-intent questions localized to each firm's home market and 2 direct brand questions. Every question ran on ChatGPT, Claude, Gemini, Perplexity, and Grok; Google AI Overviews were sampled on 3 questions per firm and reported separately. All 3,150 raw responses were archived before scoring. A firm counts as visible on a platform if any unbranded response names it, matched against punctuation and formatting variants and verified against false positives, and recognized if the platform answers direct brand questions substantively. Scoring definitions were locked before any scan ran. Full methodology and scoring code are available on request at research@faneros.ai.