GEO for Law Firms: Optimizing for AI Search Engines

TL;DR: GEO for law firms is mostly local AI visibility work plus a compliance layer that generic guides leave out. When someone asks an assistant for a lawyer, the answer is built from your Google Business Profile, state bar listings, legal directories and reviews rather than from your blog. Every tactic also has to clear your state's advertising rules, and ABA Formal Opinion 512 makes you responsible for the accuracy of anything AI drafted for you.

This article is general information about marketing practice, not legal advice. Advertising rules vary by state and your bar's rules govern.

Key Takeaways

• Start with entity data, not content. Bar listings, Google Business Profile, Avvo, Justia and Martindale are what assistants draw on for referral questions. Consistency across them matters more than publishing volume.

• Local AI recommendations are a narrow field. SOCi's 2026 index of over 350,000 locations found ChatGPT recommended about 1.2% of them against 35.9% appearing in Google's local pack, with roughly 45% overlap.

• Specialist claims are a bright line. ABA Model Rule 7.2 restricts stating or implying certification as a specialist. Describe practice areas rather than claiming expertise your certification does not support.

• Case results and testimonials need handling. Disclaimers that prior results do not guarantee a similar outcome are the standard approach, and several states impose specific requirements.

• You own what the AI wrote. ABA Formal Opinion 512 applies existing duties to generative AI. An AI-drafted bio or case summary is still your communication under Rule 7.1.

• A website chatbot must disclose that it is not a lawyer. Bar guidance treats client-facing AI chat as advertising, requiring a clear disclaimer that it is an AI program rather than a lawyer or firm employee.

Anchor targets assume your CMS generates slug IDs from heading text. Confirm on staging and adjust the hrefs if your platform slugifies differently.

How clients find lawyers through AI now

Legal is a local, high-stakes, heavily researched purchase, which makes the AI referral path specific enough to plan around.

What actually happens

Someone describes a problem rather than naming a practice area. They ask about a situation, the assistant explains the legal concept, and then they ask who can help. That second question is the one that matters, and the answer is assembled from business profiles, legal directories, review platforms and local best-of lists rather than from your firm's blog.

Assistants also do not search the phrase "near me". They infer location from IP address or the conversation and rewrite the query into a city name before searching. Optimise for "best [practice area] attorney in [city]" rather than for proximity.

The two question types

Both matter and they are won differently. Informational questions are where your content earns citations and builds the corroboration that supports the referral answer. Referral questions are won on entity data.

Most firms invest entirely in the first and neglect the second, which is backwards for a practice that needs the phone to ring.

The honest scale

SOCi's 2026 Local Visibility Index, covering more than 350,000 business locations, found ChatGPT recommended roughly 1.2% of them, against 35.9% appearing in Google's local pack, with only about 45% overlap between the two groups.

Two things follow. The field is narrow, so being named is genuinely valuable. And ranking well on Google Maps does not mean an assistant will recommend you, which is why firms with strong traditional rankings find themselves absent from AI answers and cannot explain why.

What this means for a small firm

The work is mostly free and mostly administrative: complete your profiles, claim your directory listings, make every detail consistent, build recent reviews, and publish genuinely useful jurisdiction-specific answers. A solo practitioner can do this. It does not require a marketing department or a large budget, which is why it is worth doing properly before considering software.

The compliance layer

Every tactic in this article has to clear your state's advertising rules. Generic GEO guidance does not account for this, which is why a legal marketing programme built from one can create problems a general business would never face.

General information, not legal advice. Your state's rules govern and they vary meaningfully.

The three rules that shape everything

All fifty states have adopted versions of these, and the variations matter. Some states require specific disclaimers that others do not, and some have filing or retention requirements for advertising materials.

Where GEO specifically touches the rules

The payment distinction worth understanding

Model Rule 7.2(b) generally prohibits giving anything of value to a person for recommending your services, while permitting payment of the reasonable costs of advertisements.

That distinction decides how you can buy marketing. Paying a platform for advertising is ordinarily permissible; arrangements that function as payment for referrals are a different question. If you are evaluating pay-per-lead or pay-per-case models, that is a conversation for your ethics counsel rather than your marketing agency.

The practical approach

Run your GEO plan past whoever handles your firm's ethics compliance before executing it, not after. Most of the work in this article is uncontroversial. The three places that need review are case results, testimonials and anything that speaks to prospective clients in your firm's voice, including a chatbot.

Build the disclaimer and review step into the workflow once and it stops being a recurring question.

Entity and directory foundations

This is the highest-return work for a law firm and almost all of it is free. Referral questions are answered from these sources, so inconsistency here is what keeps firms out of AI recommendations.

The listings that matter, in priority order

What to make consistent

One canonical version of each fact, used everywhere:

Tracking numbers are the quiet problem. Different phone numbers across directories is exactly the inconsistency that makes models hedge, and firms create it deliberately for attribution. If you use call tracking, keep one consistent primary number in the structured fields.

Reviews

Volume and recency both matter for local AI recommendations. Build a habit of asking every satisfied client, in whatever way your state's rules permit.

Check your state's position on soliciting reviews and on displaying them before building a process around it, since this is one of the areas where state rules diverge most.

The audit

Build a grid: attributes down the side, directories across the top. Score each cell against your canonical version. Most firms score below 60% on first measurement, and closing that gap is a few hours of data entry with no development work, no approvals and no budget.

That is the single best use of a first afternoon on this.

Attorney bios as entity anchors

In legal, the entity that matters is often the individual lawyer rather than the firm. Clients ask for a person, and assistants connect an attorney to practice areas, jurisdictions and credentials. Bio pages are where that connection is made or lost.

What a retrievable bio contains

The opening sentence

Most attorney bios open with a narrative paragraph. Replace it with a direct sentence stating who the lawyer helps, with what, and where.

That sentence is what an assistant can lift when answering a referral question. A bio opening with a story about a childhood interest in justice has nothing extractable in its first fifty words.

The specialist trap

This is where legal diverges sharply from other verticals. Model Rule 7.2 restricts stating or implying certification as a specialist, and the words that work best for retrieval are often the words the rule restricts.

The descriptive version is also more retrievable, because it states what the lawyer actually does rather than making a claim a model cannot verify.

Check your state's specific rules, since some permit qualifying language that others do not.

Schema for bios

Person schema on each bio, linked to the firm's LegalService or Organization markup, with jobTitle, alumniOf, knowsAbout and memberOf populated accurately. Validate it with the Rich Results Test.

Keep knowsAbout descriptive of practice areas rather than phrased as expertise claims, so your structured data says the same thing as your visible copy.

Practice area and jurisdiction content

Legal content has a structural advantage that most firms ignore: law is jurisdictional. The same question has different answers in different states, and that specificity is exactly what assistants need and what generic content cannot provide.

The strategy

Stop writing general explanations of legal concepts. Every firm in the country has published those, they are near-identical, and they compete with each other for the same passage.

Write the jurisdiction-specific version instead.

The right column is narrower, has fewer competitors, matches how people actually ask, and only a firm practising there can write it credibly.

The format that gets quoted

The maintenance obligation

This is heavier in legal than anywhere else. A page stating a filing deadline that changed is not merely stale, it is a potentially misleading communication under Rule 7.1 and it could harm a reader relying on it.

Schedule a review cycle by practice area, assign an owner, and show a visible last-reviewed date. Budget roughly 15% of your content time for maintenance rather than new pages.

Where to start

List the twenty questions clients actually ask in intake calls, in their words rather than in legal terminology. Those are your first twenty pages, and they are better targets than any keyword tool will produce because they come from real matters.

That list is also the best use of your intake staff's knowledge, and it costs nothing to gather.

Case results and testimonials

These are powerful corroboration signals and the most regulated content on a law firm website. Both are generally publishable and both need handling.

General information, not legal advice. Your state's rules govern.

Case results

The standard approach is a clear disclaimer that prior results do not guarantee or predict a similar outcome. Several states impose specific requirements about wording and placement, and some require more than the standard formulation.

An important nuance for GEO: an assistant may quote a result without carrying your disclaimer. That is outside your control and it is a reason to write results conservatively, so that the sentence remains accurate and non-misleading when read on its own.

That is a useful test generally. If a passage would be misleading stripped of its surrounding context, rewrite it, because retrieval strips context by design.

Testimonials and reviews

Under the ABA model rules truthful testimonials are generally permitted, and the comments no longer treat client endorsements as inherently likely to mislead. Several states require specific disclaimers, and practice varies enough that this needs checking rather than assuming.

That last row causes real problems. Responding to a negative review by correcting the facts of the representation can disclose confidential information. Keep responses brief and non-substantive.

What this means for AI visibility

Review volume and recency genuinely influence local AI recommendations, so this is worth doing well. Build a compliant process once: when to ask, how to ask, what disclaimer appears, who responds and in what terms.

A documented process reviewed by your ethics counsel converts a recurring risk into a routine, and it is the only part of this article worth paying a professional to review in detail.

AI-drafted content and your duty to verify

If you use AI tools to produce marketing content, the output is still your communication. This is settled guidance rather than an open question.

What Formal Opinion 512 established

The ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512 in July 2024, its first formal guidance on generative AI. It approaches AI through the duties lawyers already hold: competence, confidentiality, communication and supervisory responsibility. The ABA published further guidance on responsible AI use in firms in January 2026.

For marketing the relevant principle is simple. An AI-drafted bio, case summary or practice page does not get a pass on accuracy because a model wrote it. It is your communication under Rule 7.1.

The practical workflow

The specific failure mode to watch

Generative models produce confident, plausible, incorrect legal citations. Courts have sanctioned lawyers for filings containing fabricated case citations, and the same failure mode appears in marketing content where it is less likely to be caught.

A blog post citing a non-existent case or misstating a statute is a misleading communication about your services and it damages exactly the credibility the content was meant to build. Verify every citation against the primary source before publishing. Every one.

Confidentiality

Do not put client information into general-purpose AI tools when producing marketing content. Drafting a case study from matter details in a consumer chatbot is a confidentiality question before it is a marketing one.

Use anonymised, client-approved facts, and confirm what your firm's AI use policy permits.

The honest assessment

AI drafting genuinely helps law firms produce content at a pace they otherwise could not. The editing requirement is heavier here than in any other vertical, because a wrong fact carries professional consequences rather than merely commercial ones.

Budget more review time than a general business would, and treat the lawyer review step as non-negotiable rather than as a bottleneck to optimise away.

Chatbots and the disclosure requirement

AI chat widgets are commonly bundled with marketing platforms and sold to law firms as after-hours intake. They can work, and in legal they carry a requirement most vendors will not mention.

The disclosure

Bar guidance on AI treats client-facing chatbots as subject to lawyer advertising restrictions, including a clear disclaimer that the chatbot is an AI program rather than a lawyer or an employee of the firm.

Put that disclosure where a user sees it before interacting, not in a terms link. If a prospective client could reasonably believe they are talking to someone at your firm, the disclosure has failed its purpose.

The other risks

What a compliant setup looks like

1. Clear AI disclosure before the first exchange.

2. A statement that using the chat does not create an attorney-client relationship.

3. Scope limited to scheduling, directions, practice areas and general information.

4. An explicit refusal pattern for specific legal questions, routing to a consultation.

5. Jurisdiction qualification early in the conversation.

6. Intake data flowing into your conflicts check before anyone responds substantively.

7. Transcripts retained according to your record policy.

The honest recommendation

For most firms a well-configured contact form with clear expectations about response time is lower risk and converts comparably. A chatbot earns its place when after-hours enquiry volume is genuinely high and the configuration above is actually implemented.

If you deploy one, have your ethics counsel review the configuration and the disclosure language before it goes live, not after. This is the single highest-risk item in a law firm's GEO programme, and it is the one most often switched on by a marketing team without review.

The technical layer

Smaller than vendors suggest and worth getting right once. Three things matter and the rest is noise.

1. Crawler access

Verify in server logs which AI user agents reached your site and what status they received. Your robots.txt states intent; your hosting, WAF or security plugin determines outcome.

Training and retrieval crawlers are controlled separately. Blocking training bots costs nothing in AI search visibility; blocking retrieval bots removes you from AI answers. Many firms blocked everything on advice given in 2023 and never revisited it, which is worth a deliberate review.

2. Schema markup

Validate everything with the Rich Results Test. Schema helps systems confirm your details; it is not a ranking lever and it does not substitute for the entity consistency work.

3. Rendering

If your practice area pages or bios load content via JavaScript, check what a fetch without JS returns. Content that is not in the server response may not be parsed. Many law firm sites built on page builders have this problem and nobody has checked.

What you do not need

No AI.json file. No LLM-specific robots.txt variant. Google's documentation states plainly that site owners do not need to create new machine-readable files, AI text files or markup to appear in its AI features.

If a vendor's pitch leads with generating those files, that is a useful signal about the rest of their advice.

The whole technical checklist

• Server logs confirm AI user agents receiving 200 responses

• Training versus retrieval crawler decision made deliberately

• LegalService and Person schema present and validating

• Practice area and bio content present in the server response

• Core Web Vitals passing on key templates

• Google Search Console connected, including the generative AI report

Six items. For most firms that is an afternoon with a developer, and it is the end of the technical work.

Measuring it without a budget

A firm does not need an AI visibility subscription to know where it stands. Twenty minutes a month answers the question, and it works until you are tracking roughly fifty prompts.

The manual check

Open ChatGPT, Gemini and Perplexity. Ask ten questions your clients actually ask, twice each. Record whether your firm was named, how it was described, and which firms and sources appeared instead.

For a local practice the questions look like:

• Best [practice area] attorney in [your city]

• Who should I call about [specific situation] in [your city]

• [Practice area] lawyers near [neighbourhood or courthouse]

• Is [your firm name] a good [practice area] firm

• [Your firm] vs [main local competitor]

• What is [jurisdiction-specific question you have published about]

That last one tests whether your content is earning citations, which is a different question from whether your firm is being recommended.

What the answers tell you

Free tools worth connecting

All free, all first-party, and together they cover more than most paid tools would tell you about your own firm.

When to pay

Past about fifty prompts, or if you are managing several office locations or practice groups separately. Below that a spreadsheet and twenty minutes a month is sufficient and more honest than most dashboards.

The measure that matters most

Ask every new client how they found you, and record the answer. Attribution from AI referrals is genuinely difficult technically, and intake is where the real signal lives. Add "did you use an AI assistant?" to your intake questions and you will learn more in a quarter than any tool will tell you.

A 90-day plan

Written for a firm without a marketing department. Roughly four hours a month, and the first month costs nothing.

Days 1 to 30: measure and fix the free things

Days 31 to 60: bios, content and compliance

Days 61 to 90: reviews, process and re-measure

What you will have at day 90

Consistent entity data across every source assistants use, bios that can be quoted, six jurisdiction-specific pages, verified crawler access, a compliant review process and a measured before-and-after. Total external cost: nothing.

Honest expectation setting

Ninety days of this will not necessarily put you in AI recommendations in a competitive practice area in a major city. It will put your foundations in a state where the content work that follows can actually earn citations.

In a less contested practice area or a smaller market, you may see movement inside the first quarter. Either way the entity work is prerequisite rather than optional, and it is the part most firms skip on their way to starting a blog.

Seven mistakes law firms make

1. Starting a blog before fixing the profiles. Referral questions are answered from business profiles, bar listings, directories and reviews. A firm publishing weekly while its Avvo profile shows a decade-old address has the sequence backwards.

2. Publishing generic legal explainers. Every firm has written what is a statute of limitations. Those pages are near-identical and compete with each other. The jurisdiction-specific version has fewer competitors and only you can write it credibly.

3. Using specialist language without the certification. Model Rule 7.2 draws a line here, and the words that feel strongest for marketing are often the restricted ones. Describe what you do instead.

4. Publishing AI-drafted content without verification. Formal Opinion 512 makes accuracy yours regardless of who drafted it. Models produce confident, wrong citations, and in marketing content nobody is checking them against a docket.

5. Deploying a chatbot without the AI disclosure. Bar guidance treats client-facing AI chat as advertising requiring a clear disclaimer that it is not a lawyer. This gets switched on by marketing teams routinely without review.

6. Inconsistent tracking numbers. Firms deliberately create the exact data inconsistency that makes models hedge, for attribution reasons. Keep one consistent primary number in structured fields.

7. Letting content go stale. In legal a page stating a changed deadline is not just outdated, it is potentially misleading under Rule 7.1 and it can harm a reader relying on it. Budget 15% of content time for maintenance.

The pattern

Three of seven are compliance issues that a generic GEO guide would never raise, and all three are easy to walk into by following general marketing advice.

That is the case for running your plan past whoever handles your firm's ethics compliance before executing it. Most of this work needs no review at all; three specific areas do.

The one to fix first

Mistake one. The entity and directory work is free, takes an afternoon, needs no approval and is what referral questions are actually answered from. Most firms have not finished it and are investing in content instead.

FAQs

What is GEO for law firms?

Generative engine optimisation for legal practices: making your firm retrievable and quotable so AI assistants name you when someone asks for a lawyer. For law firms it is mostly local entity work across bar listings, directories and your Google Business Profile, plus jurisdiction-specific content, within your state's advertising rules.

How do AI assistants decide which lawyers to recommend?

For referral questions they assemble answers from business profiles, state bar listings, legal directories like Avvo and Justia, review platforms and local best-of lists. Your website content supports this by earning citations on informational questions, but it is rarely what answers the referral question itself.

Do bar advertising rules apply to AI search optimisation?

Yes. Model Rules 7.1 to 7.3, as adopted in your state, govern communications about your services regardless of the channel. The areas needing attention are specialist claims, case results, testimonials, AI-drafted content and any chatbot speaking to prospective clients.

Am I responsible for content an AI tool wrote for my firm?

Yes. ABA Formal Opinion 512 applies existing duties to generative AI, and the lawyer remains responsible for accuracy. An AI-drafted bio or case summary is your communication under Rule 7.1. Verify every legal proposition and firm fact before publishing.

Can my law firm use an AI chatbot on its website?

Generally yes, with care. Bar guidance treats client-facing AI chat as advertising and calls for a clear disclaimer that the chatbot is an AI program rather than a lawyer or firm employee. Constrain it to scheduling and general information, and have the configuration reviewed.

Can I publish case results?

Usually, with a disclaimer that prior results do not guarantee a similar outcome. Several states have specific wording or placement requirements. Write results conservatively, because an assistant may quote the sentence without carrying your disclaimer with it.

How long does GEO take to work for a law firm?

Profile and directory fixes can affect answers within weeks. Content and review velocity compound over quarters. Expect three to six months before a fair assessment, and longer in competitive practice areas in large markets.

Do I need to buy software?

Usually not at first. Google Business Profile, Search Console, Analytics and manual prompt checking cover a solo or small firm properly and cost nothing. Consider paid tracking past about fifty prompts or when managing several locations or practice groups.

Getting your firm into AI recommendations

Run the manual check first. Ten questions your clients ask, across three assistants, twenty minutes. That tells you whether you have a problem and which kind, and it costs nothing.

Then fix the entity layer before writing anything. Google Business Profile, state bar listing, Avvo, Justia, Martindale, Bing Places, Apple Business Connect, all consistent against one canonical version of your firm's details. That is the afternoon that does more than a quarter of blogging.

When a firm should buy nothing

• Your directory profiles are not complete and consistent. Finish that first. It is free and it is what referral questions are answered from.

• You have not run the manual check. Twenty minutes tells you where you stand.

• You are a solo or small firm under fifty tracked prompts. A spreadsheet is sufficient and more honest than most dashboards.

• Nobody can review AI-drafted content properly. Without a verification step, content production creates risk rather than visibility.

Where tooling helps

Once the foundations are in place and the constraint is producing jurisdiction-specific content faster than your attorneys can write it, a content platform earns its place. Blazly SEO covers research, planning, drafting and direct publishing from about $39 a month.

For a law firm, build the attorney review step into the workflow explicitly. Formal Opinion 512 makes accuracy your responsibility regardless of what drafted it, and in this vertical that review is not a bottleneck to optimise away.

Summary

GEO for law firms is local AI visibility work with a compliance layer on top. Referral questions are answered from business profiles, bar listings, legal directories and reviews rather than from your blog, so fix those first and make every detail consistent. Write jurisdiction-specific answers rather than generic legal explainers, since only a firm practising there can write them credibly and they face far less competition. Keep specialist language inside what your certifications support, carry the disclaimers your state expects on case results and testimonials, and verify everything an AI drafts because the accuracy is yours under Rule 7.1. If you deploy a chatbot, disclose clearly that it is not a lawyer. And ask every new client how they found you, because intake is where the real attribution signal lives.

This article is general information about marketing practice, not legal advice. Your state's rules of professional conduct govern and vary meaningfully. Consult your ethics counsel before implementing.