Three questions that dissolve most AI-visibility claims.
We surveyed the agencies selling AI visibility to law firms. The category is sophisticated, proprietary platforms, branded scores, per-practice-area playbooks. What is almost entirely absent is a published method. Ask any vendor, including us, these three questions in writing.
A growth percentage without a starting point is unfalsifiable. A firm that appeared in one answer and now appears in two has grown 100%. That is a real sentence someone could publish honestly and it would tell you nothing.
A visibility score means nothing without the sample behind it. If the question set is not disclosed, it can be selected after the fact, and a score built from questions chosen to flatter is not a measurement.
AI answers change, and they vary between runs. An undated figure is not reproducible even in principle, and a number nobody can reproduce is not something a regulated advertiser should repeat.
Across roughly forty agencies in this and adjacent verticals, the number publishing an auditable AI-visibility measurement methodology was approximately zero. What exists instead falls into three shapes: percentage growth in AI referral traffic with no disclosed baseline; aggregate claims across a client base with no sample, no date and no protocol; and branded proprietary scores whose method is never shown.
There is also a genre worth knowing about, the “top law firm AI agencies” ranking published by an agency that places itself first. We found several, including ones whose ranked lists attribute different client rosters to the same agency depending on which version you read. Those pages are marketing, not market research.
The advertising rules bind you, not your agency.
This is the asymmetry most agencies never mention, and it is the whole reason we build the way we do.
ABA Model Rule 7.1, adopted with variations in every state, prohibits false or misleading communications about a lawyer’s services, including communications likely to create unjustified expectations about the results a lawyer can achieve.
That rule constrains your firm. It does not directly constrain a marketing vendor. Which means the risk runs one way: a vendor can publish an unverifiable claim with no consequence, and the exposure lands on the firm that repeats it. The same is true of the content a vendor ghostwrites for your site — it is your regulated communication regardless of who typed it.
So “we’ll make you the most-cited firm in your market” is more than a soft claim. It is a sentence your firm probably should not be amplifying, and one we will not write.
This describes how Mindflow works. It is not legal advice, and jurisdictional rules vary, including on testimonials, where some states permit what others restrict. Your ethics counsel governs.
Firms where the client researches before they call.
Practice area matters less than the buying pattern. Where someone reads, compares and asks an assistant before picking up the phone, being present in that research decides whether your firm gets considered at all.
Personal injury, family, criminal defence, immigration, high urgency, high research, and the most contested search market of any vertical.
Corporate, employment, IP, real estate, longer cycles, fewer buyers, and a decision driven by demonstrated expertise rather than proximity.
Estate planning, probate, elder law, a patient, careful buyer who reads a great deal before choosing anyone.
Several markets, several practice areas, where entity consistency across offices is the hardest problem and the biggest lever.
How the system works across every vertical: Maps, SEO, AI answers, conversion, reputation.
The Entity Integration Framework, Measure, Fix, Build, Prove. Written down, not asserted.
Where we are not a fit: Firms that want outcome guarantees, volume lead-buying, or content that promises results. Those are declined at the first call. If the brief is “get us more cases by any means”, there are agencies for that and we are not one.
One report. Three numbers. Zero mystery.
The same three numbers every month, defined the same way each time, each one answering the three questions above before you have to ask them.
Your presence across the searches that lead to matters, measured against five named firms agreed at onboarding. The same five every month, restated in every report. The denominator never moves quietly.
Enquiries matching the matters you actually want, defined with you at kickoff against your own intake criteria, then held constant.
Twelve real client questions, asked across ChatGPT, AI Overviews and Perplexity, with screenshots and dates. You appear in X of 12. The question set is printed in your report and does not change without telling you.
Why X/12 and not a percentage: a fraction carries its own sample size. Twelve questions is a small, honest sample, and saying so is the point, a percentage implies a precision that sampled AI measurement does not have. A number that hides its sample is the first of the three questions failing.
Alongside the three numbers, every task completed that month is listed on the Work Ledger. See a full sample report →
We’d be failing our own test if we skipped this.
A page that spends its length asking you to demand baselines and question sets cannot then present a number of its own without them. So: legal is a newer bench for us, and we do not have a published before-and-after AI-citation result for a law firm. We are not going to import one from another vertical and let the layout imply otherwise.
What we can put in front of you is inspectable: the six-layer audit, the measurement protocol with its question set and competitive set written down, and our own visibility tracked in public where you can check it against us. If the method doesn’t survive your reading of it, that is a real answer and it costs you nothing to find out.
There is also a fair counterpoint to make. Some agencies in this category are sophisticated, and one publishes real per-client citation metrics. If a firm wants a vendor with a legal-specific track record today, that is a defensible choice and we would rather say so than pretend otherwise. What we would still ask, of them and of us, is the three questions above.
Fair questions from managing partners
Every agency claims AI visibility results. How do I tell which numbers are real?
Ask three things of any number, ours included. What was the baseline before the work started? What exact questions were asked, and how many? On what date, on which platforms? A growth percentage with no baseline is unfalsifiable. An aggregate claim across a client base with no sample set and no date is an assertion. If a vendor can’t answer all three in writing, the number isn’t evidence, and it isn’t something your firm should repeat.
Can our firm repeat your numbers in our own advertising?
That is exactly the right question to put to any marketing vendor. ABA Model Rule 7.1, adopted with variations in every state, prohibits false or misleading communications about a lawyer’s services, including those likely to create unjustified expectations about results. Those rules bind your firm, not your vendor, so an unverifiable agency statistic becomes your compliance problem the moment it appears on your site. Every number we report carries its date, question set and competitive set, so it can be substantiated. Your ethics counsel still has the final word.
What about client testimonials and reviews?
Testimonial rules are state-variable and stricter than most agencies treat them. Some jurisdictions permit client testimonials under defined conditions; others are considerably more restrictive, and at least one state ethics opinion issued in 2025 holds that lawyers cannot pay for endorsements or testimonials, and limits comparative claims outside legitimate, rigorously evaluated awards. We don’t run incentivised review programmes in any vertical, so this changes little about how we work, we follow your jurisdiction’s rule and your ethics counsel’s reading of it, not a national template.
Do you work with a competing firm in my market?
No, one firm per practice area, per market. That’s a written term in the engagement, not a handshake. If we’re already running visibility for a firm in your practice area in your market, you’ll hear it on the first call and we won’t take the work. Availability is checked before anything else gets discussed.
What does this cost, and what is the term?
Monthly partnerships begin at $3,500. Multi-office, multi-practice-area and highly contested engagements typically run $8,500–$15,000+ per month. The term is 90 days, then month-to-month. That last part is worth comparing against your other quotes: fixed twelve- and twenty-four-month terms are common in legal marketing, and a long term transfers the risk of the work not performing from the agency to the firm. Full pricing →
Do you have law firm case studies?
Not with a published AI-citation result, and we won’t borrow one from another vertical to imply otherwise, a page that asks you to demand baselines cannot then skip its own. What we can show is the method: the published six-layer audit, the measurement protocol with its question set and competitive set, and our own visibility measured in public. And the fair counterpoint: some agencies in this vertical are sophisticated and one publishes real per-client citation metrics, so if a legal-specific track record today is what matters most, that’s a defensible choice.
Who writes the content, and who reviews it?
We draft; an attorney at the firm reviews and is named as the author or reviewer. That isn’t a formality. A named, credentialed author is one of the strongest signals an answer engine has for deciding whether to repeat what a page says on a legal topic, and it is also what makes the content defensible as your firm’s own regulated communication.
Who does the work?
Youssef, the founder, runs every engagement personally, with specialist support under his review. No account managers, no handoffs. Meet Youssef →
Priced by complexity. Short term by design.
There are no tiers to choose between. What a firm pays follows the real shape of the work, how many offices, how many practice areas, how contested the market. And the term stays short on purpose.
Not sure where you sit? Start with a diagnostic. The Local Visibility Audit ($3,500–$7,500) suits a single-office firm; the Strategic Search and AI Visibility Assessment ($7,500–$15,000+) covers multi-office firms and contested markets. Both are yours to keep, and both are fully credited against your first invoice if a monthly partnership begins within 30 days of delivery.
The work created for your firm remains yours.
Find out which firm AI recommends instead of yours.
Start with the free Visibility Check, 12 real client questions for your practice area, your X/12 score, screenshots, and the first worthwhile fix. No sales call.
Get my free Visibility Check