Short answer: generative engine optimization for law firms means making your firm easy for AI tools such as ChatGPT, Gemini, Perplexity and Google AI Overviews to identify, trust and cite. It rests on ordinary SEO, plus four extra jobs: consistent entity data, pages that answer questions in quotable blocks, mentions on third-party sources those tools read, and deliberate crawler access decisions.
A managing partner hears it at a bar lunch: a prospective client said she asked an AI assistant which estate planning firms in town were worth calling, and the firm across the street came up. Then an agency email arrives promising “GEO” packages. This guide sets out what generative engine optimization for law firms actually involves as a working program, what it shares with the SEO you already pay for, what is genuinely new, how to measure it, and how to judge anyone selling it.
GEO, AEO and SEO: three names, one foundation
Search engine optimization is the work of earning a position in a list of links. Answer engine optimization (AEO) is an older label for shaping content so a system lifts it directly as an answer, which started with featured snippets and voice assistants. Generative engine optimization (GEO) is the newest label: influencing what a large language model writes when it composes an answer from several sources, and whether your firm is named or linked inside that answer.
The output differs: a search result offers ten choices, a generated answer offers one paragraph and a few citations. The inputs barely differ. Assistants that search the live web rely on a crawl and an index, and they mostly retrieve pages that already rank and say something specific. So GEO is not a replacement for SEO. It is SEO with a different scoreboard and more weight on your reputation outside your own site.
How an AI answer decides which firms and pages to use
Nobody outside these companies knows the exact selection logic, and it changes. Testing shows a pattern stable enough to plan around: for current or local questions, the assistant searches, pulls a set of pages, and writes from them. Pages that make the set tend to be crawlable (text in the HTML the server sends, not behind click-to-load tabs), well structured (a question heading with the answer directly under it), entity-consistent (firm name, address, phone and lawyers match the rest of the web), and corroborated (the firm also appears in directories, bar listings, local news and reviews).
Legal questions get extra caution, which is why assistants often point to directories or bar referral services instead of naming a firm. Our post on whether ChatGPT recommends your law firm walks through a ten-minute test that shows where you stand today, and our guide to AI Overviews for law firms covers which legal queries trigger Google’s summaries. The rest of this post is the program that sits behind both.
Entity work: making the firm impossible to confuse
An entity is a thing a machine can identify with confidence: your firm, each lawyer, your office. Language models can merge similar names, so a hypothetical “Harper Law” in a city that also has a “Harper and Lane” risks being blended into a composite firm that does not exist. Entity work prevents that.
Pick one canonical firm name, street address and phone number, and make every listing match it exactly: site footer, contact page, Google Business Profile, Apple Business Connect, Avvo, Justia and your bar profile. If the firm rebranded or merged, say so on the about page with both names, so an old article can be connected to the current firm.
Then make the same facts machine-readable. Organization or LegalService markup on the site describes the firm; Person markup on each lawyer’s page describes the individual, with a sameAs property pointing to that lawyer’s state bar profile, directory listings and LinkedIn. schema.org itself steers sites toward LegalService rather than its older Attorney type, so do not let a plugin output both and create two competing descriptions of one business. Our guide to schema markup for law firms shows the setup without duplicate entities. Finally, each lawyer needs a real bio page stating bar admissions, jurisdictions and the work they actually do; the attorney bio page guide covers what to write.
Writing pages an answer engine can lift
Generated answers are assembled from passages, and a passage gets used when it is short, self-contained and specific. Most law firm pages bury the answer in paragraph five, wrapped in reassurance.
The fix is a page pattern. Open each practice page with a two or three sentence block that defines the matter in your state and says who the firm helps. Write section headings as the questions clients type (“Can my spouse keep the house if the mortgage is only in my name?”), with the plain answer directly beneath. State definitions once per practice and per state, because probate in Ohio and probate in Florida have different correct answers.
Distinctive material does the most work: a county table of where filings are heard, a plain account of how your consultations run, an honest page on fee structures. Assistants cite what is scarce. Someone at the firm must own and re-check any stated deadline, because an outdated quotable page gets quoted outdated.
Earning mentions on the sources AI systems read
This is the part of GEO that looks least like website work and matters most. When an assistant decides whether to name a firm for a recommendation question, independent sources carry more weight than anything the firm says about itself. The sources that show up repeatedly in legal answers are predictable: legal directories, state and local bar association pages, review platforms, local news, legal trade publications, law school and CLE pages, and community organizations.
Treat it as a standing assignment. Complete every directory profile with the canonical details and a real practice description. Speak at a bar section meeting and confirm the event page names you and the firm. Give a local reporter a quotable explanation when a law in your field changes. Each produces a page, written by someone else, tying your name to a subject and a place.
Reviews count too, since recommendation answers often draw on them. Ask satisfied clients, follow your state’s rules, and remember ABA Formal Opinion 496 (2021): a reply to online criticism cannot reveal confidential client information.
Crawler access: robots.txt, AI bots and llms.txt
Your robots.txt file tells crawlers what they may fetch, and the AI companies publish named user agents you can allow or block. GPTBot is OpenAI’s crawler, PerplexityBot is Perplexity’s, and Google-Extended is a token Google provides to control whether its crawled content is used for its Gemini models. Google has said Google-Extended does not change how a site appears in Search, and AI Overviews draw on the ordinary Search crawl, so blocking Google-Extended is not a way to stay out of Overviews. Several companies also run separate agents for live search and user-requested fetches, so check each vendor’s current documentation before writing rules.
Allowing or blocking them is a business decision. Objecting to training commercial models on your articles is legitimate, but blocking a tool’s crawlers means that tool cannot read you when it searches. For firms seeking clients, allowing search and retrieval agents is the sensible default, with training decided separately where vendors offer separate controls. Also check that a security plugin or CDN firewall is not blocking these agents by accident.
llms.txt is a proposed root file that summarizes a site’s key pages for language models. It is an emerging convention, not a standard, with no public evidence that major assistants rely on it. Adding one is cheap and harmless; paying for it as a ranking lever is not.
The GEO program on one page
Here is the whole program as a working checklist your team or vendor can be held to. Every row has a check you can run yourself without special software.
| GEO lever | What to do | How to check it |
|---|---|---|
| Entity consistency | One canonical firm name, address and phone across site, Google Business Profile, Apple Business Connect, bar and legal directories | Search the firm name with the city and compare the first two pages of results against your canonical version |
| Structured data | LegalService for the firm, Person for each lawyer with sameAs links to bar profiles | Run key pages through Google’s Rich Results Test and the Schema Markup Validator; confirm one firm entity, not two |
| Answer-shaped pages | Opening answer block, question headings, one clean definition per practice and state | Read only the first 60 words of each top page: do they answer the heading? |
| Third-party mentions | Directory profiles, bar section pages, local press, trade articles, reviews | Note which sources assistants cite in your prompt tests, then check whether you appear on them |
| Crawler access | Deliberate robots.txt rules for named AI agents; firewall not blocking them by accident | Read robots.txt yourself and review server or CDN logs for those user agents |
| Rendering | Core content in server-sent HTML, not behind click-to-load tabs | View the page source and search for a key sentence; if it is missing, crawlers may miss it too |
| Measurement | Monthly prompt tests, AI referral tracking, intake source question | A dated log of prompts and answers, a GA4 report, and intake notes reviewed together |
Measuring generative engine optimization for law firms
Answers vary by run, user and day, so no single tracker tells the truth. Read three imperfect signals together.
First, scheduled prompt testing. Write fifteen to twenty questions in client language, mixing situations (“my father died without a will in Maricopa County, what happens to his house”) with recommendations. Run the set monthly in ChatGPT, Gemini, Perplexity, Microsoft Copilot and Google, in fresh sessions, and log which firms and sources appear. One run is noise; six months show direction.
Second, referral traffic. In GA4, assistant visits that pass a referrer appear under sources such as chatgpt.com, perplexity.ai and gemini.google.com; a custom channel group bundles them. Expect undercounting, since some apps strip the referrer and AI Overview clicks sit inside Google organic.
Third, and most important, intake. Add “how did you find us” to every call script and form, with an AI assistant option and a free-text follow up. Our guide to law firm website conversion tracking covers tying calls and forms back to their source so these answers end up in the same report as everything else.
How to evaluate a GEO agency or AI search optimization company
First test: anyone who guarantees AI citations is promising something nobody controls. There is no paid placement inside a generated answer. A credible vendor talks about inputs they control and signals they can show you.
Ask to see their prompt-testing method and a sample log. Ask whether they will fix your directory listings or only report on them. Ask whether their site changes also help ordinary search; if not, be suspicious. Ask which lawyer reviews content before it goes live, since the firm answers for everything it publishes. And ask what they refuse to do: hidden text aimed at AI crawlers, fake reviews, mass-produced thin pages.
Be wary of a separate product that duplicates SEO you already buy. GEO usually works best as added scope on the existing SEO engagement: entity work, third-party mentions and prompt testing. That is how we approach it in our law firm SEO services, and it is the structure worth asking any provider about.
Ethics: Rule 7.1 when a machine retells your story
Generated answers sometimes get facts wrong about firms: an inflated verdict, a practice area the firm dropped years ago, a claim of being “the best” that the firm never made. You cannot fully control what a third-party tool writes, but you do control what the firm publishes, and ABA Model Rule 7.1 prohibits false or misleading communications about a lawyer’s services. Pages built to be quoted are still lawyer advertising, so results need the context your state requires, and specialist claims need to follow your state’s version of Rule 7.2.
When an assistant misstates something, trace the source, often an old press release, a stale directory profile or an ambiguous sentence on your own site, and correct it. Never plant claims you would not make in a print ad. Advertising rules vary by state, so check yours through our attorney advertising rules by state hub. This is not legal advice; check your state bar’s rules.
Frequently asked questions
Is generative engine optimization different from SEO?
It shares the same foundations: crawlable pages, clear structure and authority. The difference is the goal and the weighting. GEO aims for citations and mentions inside generated answers, which puts more weight on entity consistency and on what independent sources say about the firm.
Can any agency guarantee my firm will appear in ChatGPT answers?
No. Generated answers vary between runs and users, and no vendor can buy or reserve a place in them. Treat a guarantee as a reason to walk away.
Should a law firm block GPTBot and other AI crawlers?
It is a business choice. Blocking protects content from some uses but means those tools cannot read your pages when they search. Most firms that want new clients allow the search and retrieval agents and decide the training question separately.
Does an llms.txt file help a law firm get cited?
There is no public evidence that it does. It is a proposed convention that is cheap to add and harmless, but it should not be sold or bought as a ranking lever.
How long does GEO take to show results?
Entity cleanup and page restructuring can be done in weeks, but third-party mentions build over months. Track monthly prompt logs, AI referral traffic and intake answers, and judge direction over two quarters rather than any single test.
The short version
Generative engine optimization is SEO with a new scoreboard. Make the firm one unmistakable entity everywhere it appears, write pages whose answers can be quoted on their own, earn mentions on the directories, bar pages and local press that assistants read, make deliberate crawler decisions, and measure with prompt logs, GA4 referrals and intake questions. Ignore anyone who guarantees citations.
Want to know where your firm stands before spending on any of it? Request a free law firm SEO audit and we will review your entity data, structure and crawler settings alongside your ordinary search visibility.
