The two-tier finding
In one metro PI market we measured across four AI surfaces, the pattern was unambiguous. When a main-list firm's site was cited on a money question, that firm appeared in the answer's recommendations at close to 100%. When a tail-cohort firm's site was cited — one firm's 38 times in a single window — the firm never cracked the main recommendation list. It shows up as an honorable mention, a map card, a per-firm breakdown entry. Visible everywhere, chosen nowhere. And here's the part that stings: several of those locked-out firms were actively investing in their websites the whole time. Effort without absorption is the default state of this vertical.
What separates the tiers (from our panels)
| Signal | Main-list firms | Tail-locked firms |
|---|---|---|
| Review volume | Thousands — an order of magnitude higher | Dozens to hundreds |
| Best-of / directory layer | Present in the city best-of lists the engines cite (one such page: cited 17× in one answer set) | Thin or absent |
| Published results | Verdicts plus the corroborating stack | Verdicts alone — we measured eight-figure verdict tables that still never bought a main-list seat |
| Credential route | Board certification published in extractable, corroborated form — on reasoning engines this alone earns namings | Certifications held but buried, or corroborated nowhere off-site |
Two findings worth sitting with. First, the reasoning engines now treat board certification as a filter — we've watched answers open by explaining the certification, then recommend only certified firms, citing the state registry. A certification you hold but don't publish in liftable form is, to these systems, a certification you don't have. Second, technical hygiene can silently lock you out: we've measured a firm whose robots.txt blocked its own location pages from every crawler — including the ones that write these answers.
Built for bar advertising rules
Attorney advertising is regulated, and AI-generated marketing doesn't get an exemption. Every page we draft goes through a documented approval gate — you approve every word before it publishes, with a per-page log of who approved and when, a versioned archive, and drafting checks that flag outcome guarantees, specialization claims your state bar restricts, and testimonial rules. We never take your site credentials; your team publishes. This isn't legal advice, and your compliance review stays yours — the gate exists so that review has something to sign.
The branded check nobody runs
Ask an engine directly whether your firm is reputable and you'll usually get a sourced, specific answer — built from your bar profile, your reviews, your site, and whatever else it finds. We've measured branded answers that surfaced stale rosters and unverifiable claims alongside the accurate material. You don't get to pick which version a claimant sees. Measuring it is the only way to know.
How we measure
Six surfaces, every money question run repeatedly, every run on file. Read the method →
See your firm's numbers
A free report on your real claimant questions, in 48 hours. Get your report →
A sample for your category
See the sample report format and the relevant field study. The sample is illustrative; your report measures your own market.