GEO for law firms: getting cited by AI on legal questions

GEO for law firms rests on a simple principle: if you want a law firm to be cited when someone asks a legal question to ChatGPT, Gemini or Perplexity, your content has to meet stricter requirements than in almost any other sector. AI models treat legal information as high-risk: a wrong answer can cause real harm, so the model is cautious about which sources it cites and how. In this article you will read why legal content is weighed more strictly, how to make your expertise and authority demonstrable to an AI, and what role disclaimers play in that.

Why does AI weigh legal content more strictly?

Legal content falls under what Google calls “Your Money or Your Life”: topics where incorrect information can genuinely harm someone. Law sits in the same risk category as health and finance, because bad advice about a dismissal, a rental dispute or a liability matter has direct consequences for a person’s life.

For that category, the models and the search systems behind them apply a much higher bar when it comes to E-E-A-T: Experience, Expertise, Authoritativeness and Trustworthiness. It is not a direct ranking dial, but a framework used to assess the quality of a source. For a recipe it barely matters. For an article on the limitation period of a debt claim, it weighs heavily.

The consequence is simple but strict. If an AI doubts whether your legal content is accurate and authoritative, it would rather not cite you than take a risk with a dubious source. In an ordinary niche you can still get away with good, fluent content. In the legal niche you must additionally prove who wrote it, what it is based on and within which limits it applies.

How do you demonstrate legal E-E-A-T to an AI?

You demonstrate legal authority by making crystal clear on every page who the author is, what qualifications they hold and which sources the content relies on. An AI cannot sense expertise, it infers it from the signals it finds in and around your content.

Concretely, that means a few things for a law firm:

• Visible authorship. Put a real lawyer as the author of an article, with their role, bar or specialisation and a short biography. Anonymous or generic “editorial” content scores considerably lower on trustworthiness.
• Legal grounding. Refer to the law, the code or the case law you base yourself on. An AI recognises that as grounding: it can link your claim to a verifiable source and is then more willing to cite you. We explain how that mechanism works in grounding: how AI determines what is true.
• Recency. Legislation changes. State when an article was last reviewed and keep it up to date, because outdated legal advice is, in this category, a trustworthiness risk.
• Consistent identity. Make sure your firm name, specialisations, address and people are mentioned identically everywhere: on your site, on your profiles and in legal directories. Contradictory data weakens the picture a model has of you. You can read more about that in entity consistency and AI visibility.

The common thread: make it easy for the AI to establish that a real, qualified lawyer stands behind the content. That is exactly the kind of signal that, in a high-risk category, makes the difference between being cited or being skipped. For a broader view of what GEO is and how it works, start with our guide to generative engine optimization.

What role does the disclaimer play?

A good legal disclaimer works in your favour, not against you: it tells the AI within which limits your information applies and thereby makes your content precisely safer to cite. Many firms see a disclaimer as a mandatory formality in the footer. In a GEO context, it is a trustworthiness signal.

Models are trained to be cautious on legal topics and often to add a caveat themselves (“this is not legal advice, consult a lawyer”). Content that already neatly delineates that it concerns general information and does not replace tailored advice aligns with that cautious behaviour. You confirm what the model wants to communicate anyway, and that makes you a more comfortable source.

So write your disclaimer clearly and specifically. State that it concerns general information, that every client’s situation differs and that personal advice remains necessary. Preferably place it near the content it relates to, not only hidden at the bottom. That way both human and machine read it in the same context.

Do watch the balance, though. A page entirely locked down with caveats without ever giving a concrete, usable answer is precisely less attractive for an AI to cite. First give a clear, correct answer to the question, then frame it. Value first, caveat second.

Why is GEO for law firms an authority game, not a volume game?

For a law firm, GEO is not about as many mentions as possible, but the right mention on a high-value question. Legal searches often have lower volume than consumer topics, but the value of a single case is high. One potential client with a liability matter or an acquisition case who reaches you via an AI answer weighs more than a thousand loose visitors.

That aligns with our view of visibility: we care about leads and revenue, not vanity metrics. A mention in ChatGPT is only valuable if it moves the right client to the right firm. The same logic applies more broadly in professional services, as we describe in GEO for B2B. See for instance also GEO for accountants and accounting firms, where the same strict trustworthiness requirements around figures and taxation apply.

For firms with a geographic service area, another layer is added. Many legal questions are local or regional in nature (“employment lawyer in Ghent”), and AI models increasingly take place and region into account. How to respond to that, you can read in local SEO for LLMs.

Be honest about the limits as well. GEO is not a button that guarantees clients, and no one can promise you a fixed spot in an AI answer. What you can do is make your firm the most reliable, best-substantiated source for the legal questions you excel at. That is a long-haul effort, but in a high-risk category reliability beats tricks.

If you are mainly after the classic side, meaning how your firm shows up in the ordinary results and the local map pack of Google, read our article on SEO for lawyers. This article covers the AI layer on top of it.

How do you get named for “best lawyer in [city]” questions?

More and more people no longer ask an AI assistant for an explanation, but for a name: “which employment lawyer in Ghent do you recommend?” or “a good firm for a divorce in Hasselt”. The answer is then not an article but a short list of firms, sometimes with a reason. A firm that is not on it does not exist for that person, even if its site ranks decently in Google.

For local recommendation questions, an AI combines two kinds of signals. The first are content signals: is this firm recognisably active in this area of law, with pages and articles that show it? The second are local signals, largely the same as in Google itself. For local results, Google names three main factors: relevance (does the profile match the search), distance (how far the firm is from the searcher) and prominence (how well known the business is, including through links, mentions and reviews). Assistants that search the web read largely the same sources: your business profile, your site, bar and lawyer directories, local press and review platforms.

What that asks of a firm in practice:

• A page per area of law, with the location. An AI names you for “employment law in Ghent” when your site explicitly says that you practise employment law and where your office is. A list of ten areas on a single page does not give that signal.
• A complete and consistent business profile. Category, address, areas of law and opening hours must match your site and the bar directories. How to strengthen that profile is covered in our guide to the Google Business Profile.
• Mentions outside your own site. An opinion piece in the regional press, a talk at a professional association or a listing in a legal guide gives the model a second source that confirms what you say about yourself.
• Real reviews, not bought ones. Reviews count towards prominence. Ask for them properly, without any reward and without putting anyone under pressure.

How do you measure your share of AI answers per city and area of law?

You cannot read AI visibility in Search Console, so you measure it yourself, with a fixed list of questions. The method we use in our audits can be applied in a simple form:

1. Draw up the question list. Per area of law and per city, five to ten questions phrased the way a client asks them: “best tenancy lawyer Antwerp”, “which firm helps with a dismissal for serious cause in Bruges”, “lawyer for the acquisition of an SME in Limburg”.
2. Ask every question in several assistants. ChatGPT, Gemini, Perplexity and Google AI Mode give different answers. Use a new conversation without history.
3. Repeat the measurement. AI answers fluctuate. Ask the same question on different days and judge the pattern, never a single output.
4. Record who is named and who is cited. Per question: which firms appear in the answer, which sources are cited, and are you among them?
5. Calculate your share. Your share is the proportion of answers in which your firm appears. The reverse count matters just as much: for how many questions is another firm named and you are not?

That second count is usually the wake-up call. Our own measurements among local service providers regularly show that a firm with a solid site and good reviews barely appears in AI answers, while two or three other firms in the same city are named systematically. How to approach that measurement for your own market in Belgium is covered in how to get recommended by ChatGPT. For a first impression without manual work, start with our free AI visibility checker.

Deontology-proof AI visibility: what is allowed and what is not?

Everything you publish to become visible, including for AI, is publicity in the sense of the professional rules. GEO does not change those rules; it makes them more important, because an AI can take over your own wording and spread it further.

For lawyers at a Flemish bar, the framework is the OVB Code of Ethics for lawyers (in Dutch), section III.1.7. Publicity is allowed as long as it breaches no legal rule (art. 82) and is not misleading (art. 83). You may not try to win over another lawyer’s clients in a pending case uninvited, nor make a personalised offer for a specific case without being invited to (art. 84). You may only claim particular expertise if you can make it plausible on the basis of knowledge or experience (art. 85). On files you handled, you may mention their nature, scope and result, but without the client’s name unless the client expressly agrees, and only insofar as professional secrecy and data protection allow it (art. 86).

Note that the French-speaking and German-speaking bars follow their own code, which is stricter on this point. The code of avocats.be prohibits in article 5.5 any publicity that makes clients or cases identifiable, as well as mentioning the number of cases, results obtained or a success rate. Article 5.3 rules out comparative statements. If your firm works in both regions, or publishes in Dutch and French, stick to the strictest common denominator. A detailed explanation rule by rule is in law firm websites and the advertising rules.

What does that mean for your AI visibility?

• Aim for “best lawyer” questions, but do not call yourself that. An AI decides itself whom it recommends. Your job is to supply the facts (area of law, place, experience, authors), not the superlative.
• No comparisons with other firms, not even dressed up as a “why choose us” table.
• No cases that can be traced back to a client. Describe at most the type of file and your approach, and follow your own bar’s code on results.
• Be careful with “specialist”. Use your preferred areas of practice and show the experience behind them.
• No invented or bought reviews, and no text meant to mislead an AI. Whatever a model takes from you must be as correct as what you would say yourself.

Checklist: AI visibility for a law firm

Frequently asked questions

Can a lawyer pay to be recommended by ChatGPT?

No. There is no paid placement in the ordinary answers of ChatGPT, Gemini or Perplexity. You only influence a recommendation indirectly: with clear practice-area pages, a consistent business profile and mentions in reliable sources.

Is GEO allowed under the professional rules for lawyers?

Yes, insofar as your publications comply with your bar’s publicity rules. GEO is not a separate technique that falls outside those rules: it is about correct, informative content and consistent data. Misleading claims, comparisons and solicitation remain prohibited, whatever the channel.

Can I use cases I won to build AI visibility?

That depends on your bar. The OVB code allows the nature, scope and result of files without the client’s name, within professional secrecy and data protection. The avocats.be code prohibits results and success rates. If you work in both regions, leave them out.

How quickly will I see results in AI answers?

Changes to your site and business profile can become visible within a few weeks in assistants that search the web live. A stable place in the answers for a competitive city and area of law usually takes months of consistent publications and mentions.

The short summary

Legal content is weighed more strictly by AI because a wrong answer can cause real harm. If you want to be cited as a law firm, you must make your expertise visible with real authors, substantiate your claims with law and case law, keep your data consistent and deploy a clear disclaimer as a trustworthiness signal. Treat GEO as an authority game: aim for the right mention on high-value questions, not for volume. If you work in a completely different corner, comparable principles apply elsewhere, for example in GEO for manufacturing. If you want to know how strongly your firm already appears in AI answers today and what the first steps are, take a look at our approach to becoming findable in AI search engines.

Contact us