AEO for law firms: how to get cited when clients ask AI for a lawyer

Someone who has just been in a crash or been served papers no longer starts with a Google search. A fast-growing share of them ask ChatGPT, Gemini or Perplexity to name a lawyer, and the answer they get shapes who they trust before they visit a single website. For law firms this is the highest-stakes shift in client acquisition in a decade, because legal queries are among the most AI-saturated of any category and the people asking them are ready to hire. This is how to become the firm the model names and cites.

The moment a client asks AI for a lawyer

Someone has just been in a car accident, or been served divorce papers, or had a contract fall apart. A few years ago their first move was a Google search and a scroll through the map pack. Today, a growing share of them open ChatGPT and type a plain question: "I was rear-ended in Leeds, do I have a claim and who should I call?" The answer they get back names two or three firms, summarises what to do next, and shapes who they trust before they have visited a single website.

This is the shift that answer engine optimisation exists to address. The old game was ranking a page. The new game is being the source the model reaches for when it composes an answer. For law firms the stakes are unusually high, because legal queries are among the most AI-saturated of any category and the people asking them are ready to hire.

The numbers are not subtle. Analysis by Semrush of more than 10 million keywords found that roughly 78% of legal queries now surface a Google AI Overview, reported as the highest trigger rate of any industry vertical. That means for most searches a prospective client runs, an AI summary answers first and the ten blue links come second.

Why the buyer intent here is different

Legal is not casual browsing. Nobody researches a personal injury solicitor for fun. When an AI answer names your firm, it reaches a person with an active, expensive problem, and that shows up in the conversion data.

Ruler Analytics' 2026 legal marketing figures put the AI referral conversion rate at 8.4%, ahead of organic search at 7.3% for the same sector. Broader cross-industry work in the 5W and Semrush reporting goes further, suggesting AI-referred prospects convert at around 4.4 times the rate of standard organic visitors. The mechanism is simple: a visitor who arrives from an AI citation has already asked a specific question, already read a tailored answer, and clicked through knowing exactly why they want you.

There is a second reason the legal case is compelling. Adoption is climbing fast. In an iLawyerMarketing survey of 1,110 US consumers updated in July 2026, the share who said they would use ChatGPT to research an attorney rose from 9% in 2023 to roughly 42% in 2026, while stated Google use for the same task fell from 86.7% to 71.9% in a single year. Just over half, 50.1%, said they would use at least one AI answer engine, and 9.5% said they would use only AI, no Google, no directories, nothing else.

"If the model does not know your firm exists, neither does the client it is advising."

Curious how AI engines describe your brand right now? Get a free visibility audit and see where you stand across ChatGPT, Gemini and Perplexity.

E-E-A-T is not a slogan for lawyers, it is the filter

Google and the large language models both lean heavily on experience, expertise, authoritativeness and trust for what they classify as "your money or your life" topics. Legal sits squarely in that bucket. A model deciding which firm to name in a high-stakes answer is effectively running an E-E-A-T check, and most law firm websites give it very little to go on.

The fixable gap is usually attribution and depth. Generic service pages written in the third person tell a model nothing about who is behind the advice. Pages that carry a named, credentialed author signal far more.

• Name the lawyer. Attribute substantive content to a specific, admitted practitioner with their bar admission, years in practice and areas of focus, not to "the team".
• Show the work. Case results, reported matters, published commentary and speaking engagements are the evidence an engine uses to judge experience. Publish them in plain, extractable text.
• Answer the real question. Build pages around the exact phrasing clients use ("how long do I have to file", "what is my claim worth") rather than practice-area labels only a lawyer would search.
• Keep it current. Statutes, limitation periods and procedures change. Dated, accurate content reads as more trustworthy to both readers and models than evergreen fluff.

None of this is exotic. It is the difference between a page that asserts expertise and one that demonstrates it in a form a machine can lift and cite.

Schema that tells the engine what you are

Structured data does not directly make a model like you, but it removes ambiguity, and ambiguity is what gets firms left out. For legal practices the relevant vocabulary is well defined and underused.

Use the LegalService type for the firm and the practice, and Attorney (a subtype of LocalBusiness) for individual solicitors. Mark up name, address, phone, hours, the specific areas of law you handle, the jurisdictions and courts you appear in, and the languages you speak. Add Person schema for each lawyer, linked to their bar profile and their authored articles, so the graph connects the human, the credential and the content.

The point is disambiguation. When an engine can see, in clean structured form, that "Marlow & Reed" is a LegalService practising family law in Bristol, staffed by named, admitted solicitors, it can name you with confidence. When all it has is prose it has to guess, and models hedge rather than cite when they are unsure. FAQ and HowTo markup on your genuinely useful guides gives the engine tidy, quotable blocks it can pull straight into an answer.

Local signals still decide most legal matters

Law is overwhelmingly local. Jurisdiction is not a preference, it is a hard constraint, and AI answers to legal questions almost always resolve to "a lawyer near you who handles this". That makes local authority signals central rather than optional.

Your Google Business Profile remains a primary input, so keep the category precise, the practice areas listed, the hours accurate and the review flow steady. Consistency of your name, address and phone across the web matters more than volume: contradictory listings make an engine less certain it is talking about the same firm. And do not neglect the phone. Ruler Analytics found 56.3% of legal conversions still happen by phone call, so the AI answer that surfaces your number is doing real work even when no form is ever filled in.

Reviews carry double weight here. They feed the local ranking systems and they are exactly the kind of third-party sentiment models summarise when asked "is this firm any good". Volume, recency and specific detail all help. A review that says "handled my probate in Cardiff quickly" is far more useful to an engine matching a query than a bare five stars.

Third-party authority: where legal AEO is won

Models do not trust you because you say you are good. They triangulate across independent sources, and legal has a uniquely rich set of them. This is the single biggest lever most firms ignore.

Bar association directories and regulator listings are treated as authoritative because they are. A profile on your national or state bar, on the Law Society register, on court admission rolls, all confirm the model that you are a real, licensed practitioner, which is the first thing it needs to establish before recommending you. Established legal directories that carry peer and client review layers add a further, credible signal about standing.

Being quoted as an expert in journalism, contributing to reputable legal publications, and appearing in local news on matters in your field all build the kind of citation trail models weight heavily. The goal is a consistent picture: the same named lawyer, the same firm, the same specialisms, appearing across bar records, directories, reviews and press. When those sources agree, the engine has every reason to name you and no reason to hedge.

Measuring whether any of this is working

The hard part of legal AEO is that you cannot see it in a rank tracker. There is no position five to celebrate. Either the model names you when a client asks, or it names a competitor, and traditional analytics will not tell you which happened because the conversation occurs inside ChatGPT, Gemini or Perplexity before anyone reaches your site.

So the discipline is to test the questions your clients actually ask, across the engines they actually use, and track whether your firm appears, how it is described, and which source the model credited. Run "best employment solicitor in Manchester" and its dozens of variants, watch how often you surface, note when a directory or a competitor is cited instead, and fix the gap at the source. Measured this way, AEO stops being a mystery and becomes a scoreboard: named or not named, cited or not cited, and trending in the right direction over time.

That visibility is exactly what Stellarcast is built to give a law firm - the ability to see how the AI engines answer when a client asks for a lawyer, and to fix the reasons you are being left out.