AI visibility for
law firms.
Legal queries are among the top five intents handled by AI answer engines — and also the ones most heavily filtered for authoritativeness. Firms that win are the ones whose lawyer-authored content survives the model's E-E-A-T screening. Firms that lose are the ones with generic keyword-stuffed practice pages.
The short answer
Get cited in legal by publishing jurisdiction-specific answers under a named, bar-verifiable attorney byline. Answer engines filter legal content hard for authoritativeness, so an unsigned practice-area page is discarded before extraction ever happens. The unit that gets quoted is one question, one jurisdiction, eighty to a hundred and forty words.
How AI answers legal questions today
Legal sits inside Google's YMYL framing and every answer engine applies the equivalent. A legal question triggers two things before source selection even starts: a hedge toward general information, and a preference for sources with verifiable professional authorship — courts, statutes, bar associations, and secondary legal publishers. An unsigned practice-area page is discarded before its extractability is ever assessed.
The composed answer has a predictable shape: a general statement of the rule, a jurisdictional caveat, and a recommendation to consult a licensed attorney. There is essentially one citation slot in that structure, and it goes to whichever page states the rule most cleanly while being attributable to a named person.
Jurisdiction is the discriminator that decides everything else. 'Statute of limitations for personal injury' with no state attached is answered generically from reference sources. With a state attached, the model needs a span naming the state, the cause of action and the period together. Firms lose these queries by writing about the doctrine and mentioning the state only in the page title.
One structural note specific to this vertical: Google restricted FAQ rich results to well-known authoritative government and health sites in August 2023, so FAQPage markup on a law firm site earns no search feature whatsoever. It remains the highest-weight schema check in the rubric at four points, purely as an extraction aid. Firms that ripped it out when the rich result disappeared removed the one thing making their answers liftable.
Which legal queries convert and which are vanity
The decision-point shape in legal is jurisdiction plus matter plus action: 'employment lawyer in Austin for a non-compete', 'Delaware corporate counsel for a Series A'. Somebody typing that has a problem, a location and a budget. These are winnable precisely because they are narrow enough that no national publisher bothers with them.
Cost queries are the cheapest unclaimed ground in the vertical. 'How much does a patent attorney cost' is asked constantly and answered by almost nobody with real numbers, because firms are trained not to publish them. A page with honest ranges and the variables that move them gets cited far out of proportion to the effort it takes.
'Do I need a lawyer for X' converts if you answer it honestly, including the cases where the answer is no. The people you turn away were never clients; the ones who stay arrive pre-qualified, and the honest framing is exactly what makes the page quotable by a model trying not to overreach.
Two shapes to abandon. Pure doctrine lookups — 'what is promissory estoppel' — are research, already owned by law-school and reference sources, and produce no matter. Superlative rankings — 'best injury lawyer in Miami' — get answered cautiously from directories, because models are conservative about ranking licensed professionals. You cannot write yourself into first place there, and the attempt consumes the budget that should have gone to jurisdictional answers.
The mechanism behind the cited firms
Cooley's startup hub is structured like documentation rather than like a blog: a stable URL per topic, a definitional opening sentence, and none of the client-alert framing that makes most firm content undatable and unquotable. Each page answers one question, which is the only structure that survives chunking intact.
Gunderson Dettmer attaches a named attorney to every article, so attribution travels with the extracted span. Orrick's practice pages carry FAQ blocks, the smallest liftable unit in legal content. Fenwick's concentration on a narrow domain creates topical density, so when a pre-IPO question is asked the surrounding corpus is theirs. Wilson Sonsini's syndication places the same content on publisher domains the models already treat as legal sources.
For a four-attorney firm, three of those are directly reproducible and none of them require scale. A firm with three practice areas needs three documentation-shaped hubs, not three hundred blog posts. Byline every page with a real attorney and link to the state bar profile. Syndicate monthly with proper attribution. Brand recognition is the one input you cannot copy, and it is also the one the model checks least directly — what it checks for is a verifiable person.
What does not work for law firms
City-page doorway networks. Four hundred pages of 'personal injury lawyer in <city>' with the city name swapped are near-identical text, which collapses the claim-uniqueness check. Agencies keep selling them because they still move local map packs — a real search outcome and a complete non-event for answer engines, which have no reason to quote the four hundredth copy of a paragraph.
Bulk directory profiles. Filling out Avvo, FindLaw and Justia gives the directory a stronger page about your practice area, and the citation then goes to the directory while your firm appears as a line item inside it. One well-maintained profile is hygiene. Twelve of them is a donation.
Case-result pages as the primary content asset. Headline verdicts are advertising-regulated in most states, map to no question anyone asks an answer engine, and contain no transferable rule. They belong on the site. They do not belong at the centre of the strategy.
Ghost-written content under a 'Firm News' byline. This fails the verifiable-author check worth two points, but the real cost is the vertical's authority filter, which is looking for a person with a credential. Anonymous legal content is not so much down-weighted as never considered. Gated PDF guides fail for a related reason: a model cannot complete a form, and the material you thought worth gating is exactly the material that would have been quoted.
What good looks like for a law firm
Legal benchmarks measure attribution and specificity, not volume. A firm with twelve properly signed, jurisdiction-specific answer pages outperforms one with four hundred unsigned city pages, and the table below is the fastest way to work out which one you are.
| Signal | Invisible | Competitive | Category leader |
|---|---|---|---|
| Attorney byline | None or 'Firm Blog' | On some articles | Every page, linked to bar profile |
| Jurisdiction in the answer | Page title only | On practice pages | First sentence of every answer |
| FAQ blocks per practice area | Zero to three | Eight to twelve | Twenty or more at 80–140 words |
| Secondary-publisher syndication | None | Occasional | Monthly with canonical attribution |
| LegalService and Attorney schema | Absent | Organization only | LegalService plus Attorney on every bio |
| Published fee ranges | None | On request | Ranges and cost drivers in HTML |
Start with the byline row. It is a one-afternoon change across the whole site, it is the gate this vertical actually enforces, and every other row on the table is worth less until it is done. Jurisdiction placement is second and is equally mechanical — move the state into the first sentence of each answer.
The queries that shape your category.
A non-exhaustive sample of the questions we see buyers in this space ask ChatGPT, Claude, and Perplexity — pulled from our real-query telemetry, not guessed.
- 01
“do I need a lawyer for a small business LLC in California”
- 02
“what's the statute of limitations for personal injury in Texas”
- 03
“employment law firm specializing in startup equity”
- 04
“how much does a patent lawyer cost in 2026”
- 05
“best immigration lawyers in Miami for H-1B”
- 06
“Delaware corporate law firm for series A”
The names that keep showing up.
- CooleyDocs-grade content on their /go startup hub.
- Gunderson DettmerNamed attorney bylines on every article.
- FenwickDominates pre-IPO legal queries.
- OrrickBest-in-class FAQ schema on practice pages.
- Wilson SonsiniStrong secondary citations via JD Supra.
higher citation rate on bylined articles vs anonymous 'Firm Blog' posts.
AIRank legal vertical analysis, 2026.
How to get cited in law firms.
- 01
Byline every article with a real attorney
Every page gets a named attorney, a link to their bio, and on that bio a bar number, admission dates and a link to the state bar profile that confirms them. Add a review date next to the byline and update it when the law changes. Where a non-attorney drafts the piece, publish it under the reviewing attorney's name with an explicit 'reviewed by' line rather than a generic firm byline. Retire the 'Firm Blog' author entity entirely, including on archived posts, which are still sitting in the corpus. This is a one-afternoon change across a typical firm site and it gates the value of everything else you do.
- 02
Build practice-area FAQ hubs
One hub per practice area at a stable URL such as /practice/<area>, with twenty questions answered in 80 to 140 words each. Name the jurisdiction in the first sentence of every answer, not merely in the page title. Take the questions from intake calls and from the phrasing prospects actually use, not from a keyword tool. Mark the block up with FAQPage schema and keep the markup identical to the visible text. Link each hub from the attorney bios that cover it and from the site header. Twenty properly written answers per area outperform a hundred blog posts and are far easier to keep current.
- 03
Publish on JD Supra and Lexology
Publish on your own domain first, wait for it to be indexed, then syndicate with a canonical tag or an explicit attribution link back to the original. One article a month on a single secondary publisher beats scattering across four of them. Choose pieces that state a rule or a process rather than commentary on a recent decision, because the rule pieces stay citable for years while commentary expires in weeks. Keep the attorney byline identical across both copies so attribution resolves to the same person. The value is corroboration on a domain the models already treat as a legal source.
- 04
Crawlable attorney bios with verified bar numbers
One page per attorney at a stable URL, in HTML, carrying bar number, admission dates and jurisdictions, law school, practice areas, and outbound links to the state bar profile and any professional registry listing. Add Attorney or Person schema with sameAs pointing at those same profiles. Keep the name spelled identically everywhere on the site, including in bylines and in the schema, so entity resolution has no ambiguity to work through. Avoid rendering the bio grid from a client-side directory component. This is what turns a byline from a name into a verifiable credential, which is the specific thing this vertical is filtered on.
Our state bar restricts advertising claims. Does that block this work?
No, because almost nothing here is an advertising claim. Jurisdictional explanations, fee ranges, process descriptions and FAQ answers are informational content, and the byline requirements this work depends on are the same disclosures your bar already expects. The tactics that create bar exposure — superlatives, guaranteed outcomes, headline verdict marketing — are also the ones that do not work on answer engines. Compliance and effectiveness point the same direction here.
Is FAQ schema still worth adding after Google removed FAQ rich results?
Yes, but for a different reason than before. Since August 2023 the rich result has been restricted to well-known government and health sites, so a firm site gets no search feature from it. The markup still pairs a question with a bounded answer in a form a model can lift whole, and it is the highest-weight schema check in the 47-point rubric at four points. Ship it as an extraction aid and stop measuring it against rich-result impressions.
Does syndicating to JD Supra create a duplicate content problem?
Only if you syndicate carelessly. Publish on your own domain first, then syndicate with a canonical tag or a clear attribution link back. What you gain is the same claim appearing on a domain the models already treat as a legal publisher, which is corroboration you cannot manufacture on your own site. The risk of being outranked by your own syndication is real and is managed with attribution, not by refusing to syndicate.
We inherited 400 city landing pages. Should we delete them?
Consolidate rather than delete. Keep the pages for cities where you actually have an office or a licensed attorney and rewrite those with local specifics — court, filing rules, local counsel, address. Redirect the rest into the practice-area hub. The duplicated ones are not merely useless for citation; they dilute the internal link graph that determines whether your real answer pages are reachable in three clicks.
Can a paralegal or a marketing writer draft the content?
Draft, yes. Publish under their name, no. The byline needs to be an attorney whose credentials are verifiable — bar number, admission dates, and a link to the state bar profile on the bio page. A named reviewer with a visible review date carries the same weight as a named author, so the practical workflow is writer drafts, attorney reviews, and the attorney's name and bar link go on the page.
See your standing in law firms.
Free scan. No signup. Your AI Score, the specific queries you show up for, and the three highest-leverage fixes.
Written by
The AIRank Editorial Team
Research & editorial, AIRank
The AIRank editorial team runs the 47-point scanner, the Observer pings, and the GEO research programme every week. Writing is reviewed by the core engineers who build the Injector, Blaster, and Surgeon agents.
Last reviewed
About the team →