A new category of vendor has appeared in the last year selling “AI search visibility” to law firms: checklists of schema markup and entity signals promised to get you recommended by ChatGPT and Google’s AI answers. Some of the underlying advice is sound. Almost all of the certainty around it is not.
AI search for law firms is worth taking seriously, because a growing share of people asking legal questions get an answer without clicking anything. But the honest version of this topic is smaller and less mechanical than the pitch. This post separates what plausibly influences whether a model names your firm from what is being sold as if it were understood.
No statistics here about what percentage of legal searches are answered by AI. I could not verify one I would stand behind, and neither can the people quoting them at you.
The short answer: AI assistants mostly surface firms that are already well-represented in conventional sources: indexed content that answers the question directly, a consistent business identity across the web, and third-party mentions. There is no submission process and no verified ranking factor list. Sound structured data and genuinely useful content help; anyone selling a guaranteed method is selling certainty that does not exist.
In this article
- What is actually different about an AI answer
- Where the models appear to draw from
- What is worth doing
- What is not worth paying for
- Measuring AI search visibility without fooling yourself
- What this means for a small firm
- Frequently Asked Questions
What is actually different about an AI answer
Three things change when a question gets answered rather than listed.
There is one answer, not ten. A results page gives ten firms a chance. A generated answer names two or three, or none. The distribution of attention gets sharply narrower.
The question is longer and more specific. People type differently into an assistant: full sentences, context, follow-ups. That favours content written to answer a real question over content written to hold a keyword.
Attribution is inconsistent. Sometimes the answer cites sources you can click. Sometimes it synthesises and names nobody. You can be the reason an answer is correct and get no traffic and no credit for it.
That last point is the strategically important one, and it is uncomfortable: some of the value of this work is unmeasurable by design.

Increase Your ROI With AI (law firms)
Showing up in AI answers is one piece of a bigger question: where AI actually pays a firm back. The guide covers where it does, how to measure it, and the rules any AI written copy has to follow before it goes live.
We will email you the guide and may follow up once about your marketing. No lists, no sharing, unsubscribe any time.
Where the models appear to draw from
Being careful about the epistemics here: the systems are not documented in the way search ranking factors are, they change frequently, and much of what circulates as fact is inference from small samples. What follows is a reasonable reading, not a specification.
Assistants that browse are reading the live web, which means conventional indexing still governs whether you are reachable at all. Assistants answering from training data are reflecting a much older snapshot of the web, where longstanding presence matters more than recent publishing. Both appear to lean on sources that state things plainly and are corroborated elsewhere, which is why a firm mentioned in directories, local press and its own clear content tends to surface more than one with a thin site and no external footprint.
The practical translation is unglamorous: the things that make you legible to an AI assistant are mostly the things that already made you legible to search engines and to humans.

What is worth doing
Answer questions directly, near the top. A page that opens with a clear two-sentence answer before the throat-clearing is easier to extract from than one that builds to a conclusion. This is good writing practice regardless.
Be consistent about who and where you are. Same firm name, address and phone everywhere. Contradictory business details across directories make you a less reliable entity to summarise.
Use structured data honestly. LegalService or Attorney schema, FAQ markup where you genuinely have FAQs. This helps machines parse you. It is not a magic signal, and marking up content that does not exist on the page is a way to get penalised rather than promoted.
Publish the specific things nobody else will. Real price ranges, honest comparisons, plain answers to procedural questions. Generic content has thousands of near-identical competitors and gives a model no reason to prefer you.

Increase Your ROI With AI for Law Firms
Where AI actually pays back for law firms, how to measure it first, and the rules AI-written copy must follow.
Send me the guideGet mentioned by other people. Local press, professional directories, organisations you belong to. Corroboration across independent sources appears to matter more here than any on-page tweak.
That list should look familiar. It is largely topical authority and content strategy with different framing.
What is not worth paying for
Guaranteed placement in AI recommendations. There is no submission endpoint and no advertising product to buy your way in; anyone promising placement is either guessing or misrepresenting.
Large volumes of thin "AI-optimised" pages. The same content-farm strategy that stopped working in conventional search does not work better because the reader is a model.
Schema packages sold as a standalone product. Structured data is a supporting technique. It is worth doing properly and worth roughly what an afternoon of work costs, not a monthly retainer of its own.
Tools promising precise AI visibility scores. Model outputs vary between users, sessions and phrasings. A single number implying stable measurable rank is overstating what can be observed.

Measuring AI search visibility without fooling yourself
You can track this, but only loosely, and you should be honest about the limits.
Pick fifteen or twenty questions a real client would ask. Run them monthly across two or three assistants, from a logged-out session, and record whether your firm appears and whether the answer cites you. Watch referral traffic from assistant domains in analytics, accepting that much of the influence will never show up there. Note when you get called by someone who says an AI recommended you. Anecdotal, but it is the only end-to-end signal that exists.
What you will get is a rough directional read. Treat month-to-month wobble as noise. Treat "we never appear for any of our core questions" as signal.
What this means for a small firm
Do not reorganise your marketing around AI search for law firms, and do not ignore it either. The work that improves AI visibility is almost entirely work that improves conventional search visibility and reader experience: specific content, honest answers, consistent business data, real third-party mentions.
If someone is selling you an AI visibility retainer that duplicates SEO work you are already paying for, you are buying the same thing twice with a newer label. If you have no content and a thin web presence, no amount of schema will make an assistant recommend you. There is nothing to draw on.
The firms that will do well here are the ones publishing things worth quoting. That was already true.
Frequently Asked Questions
How do I get my law firm recommended by ChatGPT?
There is no submission process or paid placement. Assistants tend to surface firms with clear, direct content that answers the question, consistent business information across the web, and mentions from independent sources. Anyone guaranteeing placement is overpromising.
Does schema markup make an AI recommend my firm?
It helps machines parse your pages accurately, which is worth doing. It is not a switch that produces recommendations, and marking up content that is not actually on the page risks a penalty rather than a benefit.
Is AI search replacing Google for legal questions?
It is changing how some questions get answered, particularly informational ones. People with an urgent legal problem still search, still read reviews and still call. Treat AI visibility as an addition to local search and reviews, not a replacement.
Should I pay for an AI visibility service?
Look closely at what is actually being delivered. If it is content, structured data and digital PR, that is legitimate work, but check you are not already paying someone for it. If it is a score in a dashboard and a promise of placement, be sceptical.
Will AI answers reduce traffic to my website?
Plausibly, for informational queries answered in full without a click. That is a reason to make sure your commercial pages are strong, the practice-area, contact and location pages above all, and to accept that some of your content's value is now influence rather than sessions.
If you want to know whether an AI visibility proposal is selling you something you already own, get in touch. Send the scope and I will mark up what overlaps.
Related reading

ChatGPT Referral Traffic Drop: Why AI Leads Look Like Direct
Where AI-driven visitors hide in your analytics and how to spot them.

Google AI Content Guidelines: Fact-Check Before You Publish
Google now says to fact-check all AI content, even titles and alt text. What to do.

GPT-6 in ChatGPT: What Small Businesses Should Do Now
What GPT-6 means for law firms, agents and contractors, plus five marketing moves.
Alex Maldonado, Founder of Maldonado Marketing, Los Angeles. This article is general information for business owners, not legal or financial advice.

