On June 10, 2026, we asked Claude a question a divorcing parent might actually type: “Do I need a lawyer for an uncontested divorce in Texas, or can a service handle the paperwork?” The answer explained that Texas allows pro se filing, flagged the residency and waiting-period rules, and said document preparation services can handle uncontested cases without children or property disputes, while recommending an attorney when custody or significant assets are involved. No firm names in the generic answer. Then we made the query local and specific, and the engine started citing sources: state bar pages, legal aid organizations, and the handful of firms and services whose content directly answered the question.
That is the whole game of aeo legal services in one test. The engines give away the educational layer for free, then cite whoever made the decision layer legible. Most of the legal industry is still optimizing for a search results page that their clients quietly stopped using for exactly these questions, because nobody wants “how to divorce” in their search history but everybody will ask a chatbot.
Map the queries that happen before anyone calls a lawyer
Legal demand starts as a private, embarrassed, fact-pattern question: can my landlord do this, what happens if I miss a custody exchange, is a verbal contract enforceable in my state. These pre-intake queries used to scatter across forums and search results. Now they consolidate into AI conversations, and the business that answers them earns the next question in the same conversation, which is “who can help me with this near me.”
Build your query map from intake calls, not keyword tools. The questions clients ask in the first ten minutes of a consultation are the questions they asked a chatbot the night before. Fifty of those, phrased the way a scared non-lawyer phrases them, is your content plan for the year.
The phrasing detail is not cosmetic. Keyword tools return the vocabulary of people who already know the legal term (“quiet title action,” “QDRO”), while chatbot queries arrive in the vocabulary of the problem (“my ex won’t sign the retirement paperwork the court ordered”). Engines bridge the two, but they cite the source that matched the asker’s frame, which means pages titled in client language with the legal term introduced in the answer outperform pages titled in lawyer language every time we test it. Your intake staff hears the client phrasing all day; a standing note where they log the exact words of the first question each caller asks will outproduce any keyword subscription you can buy, and it costs nothing but the habit.
Notice that “legal services” is eight markets, not one
Law firms are the loudest players here but not the only ones, and competition thins fast outside them. Mediators and arbitrators. Document preparation and LDA services. Court reporters and deposition services. Process servers. Expert witness practices. Legal nurse consultants. Notaries and apostille services. Legal tech tools sold to consumers. Each of these answers different questions, and in our testing the non-firm niches are dramatically less contested: ask an engine about deposition services or apostille turnaround in a mid-size metro and it visibly strains for sources. Thin sourcing is the AEO opportunity in its purest form, because one well-built site becomes the answer.
Write the answer-first page your compliance posture allows
Each query gets a page that answers in the first 80 words, in plain language, with the jurisdiction named, then earns depth below: the statute or rule, the exceptions, the moment when general information stops being enough. The structure that wins citations is the structure bar advertising rules already push you toward, educational general information with a clear not-legal-advice disclaimer and no outcome promises. The format constraint and the regulatory constraint point the same direction, which makes aeo legal services unusual among marketing channels: the compliant version is the effective version.
Authorship is the multiplier on every page. A named attorney with bar number and practice focus, listed as author or reviewer with a visible date, turns an anonymous web page into an attributed professional answer, and the engines treat the two very differently in a category where their builders fear wrong answers. The same page also needs maintenance discipline: legal answers rot on a statutory schedule, and a page citing a filing fee or a deadline that changed last session is teaching every engine that reads it to discount your domain. An annual review pass over the answer library, tied to your jurisdiction’s legislative calendar, protects the asset. Firms treat this as a cost; it is actually a moat, because the content farms competing for these queries maintain nothing, and staleness is how the engines eventually catch them.
Make jurisdiction your moat
A generic page about power of attorney competes with every publisher in the country. “Power of attorney requirements in Ohio, witnessed or notarized?” competes with almost nobody and matches how legal questions are actually asked, because law is jurisdictional and the engines know it. State-level pages beat national ones; metro-level beats state for anything tied to courts, filing, or service. The firms winning citations right now are the ones whose URL slugs read like a county clerk’s FAQ.
Jurisdiction also solves the authority problem that plagues small firms. On a national question, the engines can cite Nolo, FindLaw, or a law school resource, and they will, every time, because those domains carry weight yours cannot match. On “how long does probate take in Cuyahoga County,” the national publishers have nothing specific and the engine must choose between the court’s own site, which rarely answers in plain language, and whoever wrote the plain-language answer with the local filing details. That contest is winnable by a three-lawyer firm this quarter. Multiply it across the forty questions your county’s filers actually ask and you have a moat that a national content farm cannot economically attack, since replicating it would require doing this work for three thousand counties.
Audit your entity record before building anything new
Ask each engine what it knows about your firm or service by name. Wrong practice areas, departed partners, defunct office locations, and confusion with similarly named firms appear constantly, and every error traces to a stale source: the state bar profile nobody updated, the Avvo listing from 2019, the old About page. Fix the sources and the answers follow within weeks. Your entity record is the foundation under every recommendation query, because an engine will not recommend a business it cannot confidently describe.
Run the audit as four identical questions across ChatGPT, Claude, Perplexity, and Gemini: what does the firm do, who are its lawyers or principals, where does it practice, and what do people say about it. Score each answer for errors and note which source each engine seems to lean on, which Perplexity in particular will show you outright in its citations. The pattern across legal clients we audit is consistent enough to predict: the practice areas are five years stale, one name is wrong, and at least one engine has blended you with a similarly named firm in another state. Every one of those errors is currently shaping referrals you never hear about, because the prospect who got a confused answer did not call to tell you.
Feed the directories and databases engines treat as ground truth
Legal is unusual in having authoritative registries the engines lean on hard: state bar directories, Avvo, Justia, FindLaw, Martindale, court-approved provider lists, mediator rosters. Complete, consistent profiles across these do more for recommendation queries than most content. For non-firm services, the equivalent is industry associations and court vendor lists. Same names, same addresses, same practice descriptions everywhere, because the engines cross-check and hedge on mismatches.
Treat the cleanup as a half-day project with a spreadsheet. List every directory where your firm appears or should, capture what each one currently says, and fix the three fields that drive entity confidence: the exact legal name, the practice description, and the people. The practice description is where most firms sabotage themselves, writing “full service” into every directory because turning away work feels expensive. To an engine, “full service” is noise; “family law limited to high-asset divorce and custody modification, Travis County” is a classifiable entity it can recommend with confidence. The narrower description wins the queries you actually want and costs you only the queries you were never going to convert.
Reviews ride along with the directory layer in legal more than firms expect. Engines read Google and Avvo review text for practice-area evidence and treat review volume as a liveness signal, while prospective clients read them for the one thing your site cannot credibly say, what you are like to work with under stress. The same peak-moment discipline that works for any service business works here, asked at resolution, never conditioned on outcome, and answered publicly within two days, with bar advertising rules kept in mind for any reply that could identify a client matter.
Earn the citations that carry trust
A quote in your metro’s legal journal, a contributed column on a bar association blog, local news commentary on a rule change in your specialty: earned mentions on third-party domains are weighted beyond anything you publish yourself. One reactive-source quote per month is an achievable cadence for any practitioner, and twelve months of them changes which sources an engine reaches for when your practice area comes up. This is where aeo legal services stops being a website project and becomes a visibility practice.
Legal gives you a citation trigger most industries lack: the law keeps changing on a public schedule. Every new statute, appellate decision, and rule amendment in your specialty is a 48-hour window when reporters need a practitioner to explain what it means for normal people, and the practitioner who has a two-paragraph plain-language explainer ready before the local newsroom writes its story gets quoted in it. Watch your specialty’s legislative calendar and advance sheets the way a financial advisor watches the Fed calendar. Three or four well-timed explainers a year, placed or even just published and shared the day the change drops, will out-earn a hundred generic blog posts, and they keep paying as the reference material engines retrieve every time someone asks “what changed.”
Measure the conversation, not the click
Track three things monthly: what engines say about your brand, whether you appear for your ten money queries, and how intake sources shift. Firms running this playbook report a tell-tale pattern: prospects arrive having already “talked through” their situation with an AI and name the firm because an answer did. Click attribution misses all of it; intake interviews catch it. Add one question to intake: did you ask an AI assistant about this before contacting us?
Expect the answer distribution to move quarter over quarter, and expect the AI-referred clients to behave differently in ways that matter to the practice. They arrive better informed about the basic process, having had the patient, judgment-free explainer conversation no human consultation has time for, and they convert faster because the chatbot already absorbed the embarrassing questions. The consultation starts at “here is my situation specifically” instead of “how does this even work.” That shift quietly changes what your intake process should do, and the firms noticing it earliest are restructuring first consultations around verification and strategy rather than education, which both shortens them and raises close rates. The engines are not just a referral channel. They are pre-processing your clients, and the practice that understands what arrives pre-processed gets an operating advantage on top of the visibility one.
Start with the entity audit this week. It costs an hour, it usually surfaces at least one error feeding every answer about you, and it tells you exactly which source to fix first.