The State of AI Search Visibility for Law Firms (2026)
For the searches that bring clients, AI cites directories — not law firms. Across 540 high-intent legal queries, each asked three times (1,620 AI answers in total), a legal directory was the first source cited 77.8% of the time (1,254 of 1,612 answers). Directories took at least 51.8% of all 18,900 citations; individual law firms were so fragmented that 254 different firm websites split a thin minority, and the most-cited single firm appeared in well under 1% of citations. We call it the directory wall.
- Across 540 high-intent legal queries run 3 times each (1,620 answers, 18,900 citations), a legal directory was the first-cited source in 77.8% of answers.
- Directories took at least 51.8% of all citations. The figure is published as a floor, not a point estimate: 36.0% of citation volume is on domains not yet classified, some of which are further directories.
- Justia was cited in 66% of answers; the 8 most-cited domains were all directories before a single law firm appeared.
- Law firms are fragmented: 254 different firm sites split the firm share, and the most-cited firm appeared just 36 times in 18,900 citations — none cited consistently.
- 34.2% of the firms AI does name have no page of their own among the sources — AI knows them from directory listings, not their own content.
- Answer Share is measured as a consistency threshold (named in ≥2 of 3 runs), not a percentage, because AI answers are non-deterministic — roughly half of firms named appear in only one of three runs.
- The full method, query panel and raw per-prompt data are published, so every figure is reproducible.
Who does AI actually cite for high-intent legal queries?
Directories. Across 1,620 answers to 540 lawyer-hiring queries, a legal directory was the first source cited 77.8% of the time, and directories took at least 51.8% of all 18,900 citations. The most-cited domains are all directories — Justia, SuperLawyers, Avvo, Lawyers.com, Expertise, BestLawyers — before a single individual law firm appears.
This matters because generative engines and AI Overviews retrieve from the same high-authority sources. When a prospective client asks an AI "who's the best injury lawyer in my city," the answer is built primarily from directory pages — the firms themselves are a footnote.
| Domain | Type | Citations | Share |
|---|---|---|---|
| justia.com | Directory | 1,834 | 9.7% |
| attorneys.superlawyers.com | Directory | 1,526 | 8.07% |
| avvo.com | Directory | 1,085 | 5.74% |
| lawyers.com | Directory | 1,082 | 5.72% |
| expertise.com | Directory | 928 | 4.91% |
| bestlawyers.com | Directory | 807 | 4.27% |
| bestlawfirms.com | Directory | 728 | 3.85% |
| lawyers.findlaw.com | Directory | 408 | 2.16% |
| reddit.com | Media | 386 | 2.04% |
| lawinfo.com | Directory | 383 | 2.03% |
How much of AI's answer is directories vs. actual firms?
Directories dominate, and we publish the split as a bound rather than a tidy number because it is the hardest thing to measure cleanly. Directories are at least 51.8% of citations; identifiable law firms are at most 44% and the true figure is lower, because 36.0% of citation volume sits on domains not yet classified — many of them further directories. A wide honest bound beats a narrow number we cannot defend.
The stronger, cleaner signal is position: a directory was the very first source cited in 77.8% of answers (1,254 of 1,612). Whatever the exact volume split, directories are what an AI reads first and trusts most.
| Source type | Share of citations | Note |
|---|---|---|
| Legal directories / aggregators | ≥51.8% | first-cited source in 77.8% of answers |
| Unclassified (firms + more directories) | 36.0% | long tail; classification incomplete by design |
| Identifiable law-firm websites | 7.5% | a floor — only positively-identified firms |
| Media / .gov / .edu / reviews | 4.7% | news, bar associations, universities |
Why can't any single firm break through?
Fragmentation. The firm share was split across 254 different law-firm websites, and the most-cited single firm appeared just 36 times across 18,900 citations. No firm has accumulated enough authority to be a consistent AI answer, so directories — which aggregate thousands of firms and carry far more referring domains — win by default.
There is a second, sharper version of the same problem: 34.2% of the firms AI does name in its answer have no page of their own among the sources it cites. The AI knows the firm from directory listings, not from anything the firm published. Closing that gap is an off-page authority play, not an on-page one.
What should a law firm do about the directory wall?
Build authority the way directories have it — corroborated across the web — and stop relying on on-page content alone. The firms that break through earn independent citations (legal press, claimed and optimized high-authority directory profiles, original data), reconcile their entity everywhere (byte-identical NAP, a rich sameAs graph), and publish genuinely quotable, sourced content. Content makes you citable; authority makes you cited.
That is the strategy InterCore runs for law firms. To see where your firm stands in AI answers today, start with a free AI visibility audit, or read how the retrieval actually works in GEO for lawyers.
Methodology
We asked Perplexity Sonar 540 high-intent queries — 6 practice areas (criminal defense, employment law, family law, immigration, medical malpractice, personal injury) across 30 US metros in 23 states, in three fixed phrasings ("top {practice} law firms in {city}", "who should I hire for a {practice} case in {city}", "best {practice} attorneys in {city}"). Each query was asked three times because AI answers are non-deterministic; that is 1,620 answers and 18,900 source citations in total. For every answer we recorded the cited source domains and classified each as a legal directory, an identifiable law firm, or other.
Two honesty rules govern the numbers. First, the directory-vs-firm split is published as a bound, because 36.0% of citation volume sits on domains we have not positively classified — we never assume an unclassified domain is a firm. Second, a firm counts as "named" only when it appears in at least two of three runs, so a single lucky run is not reported as a finding. The full query panel and the raw per-prompt data are published at the dataset link, so every figure here is reproducible. This is InterCore first-party research — we are the origin of these figures.
Frequently asked questions
See where your firm stands in AI answers
Get a free AI visibility audit — we’ll show you exactly which queries surface your firm versus a directory, and what closes the gap.
Get the free audit →