How to choose an AI visibility platform

You have been asked to pick an AI visibility platform. The category is new, every vendor sounds the same, and the partners want an answer this quarter.

The choice is harder than it looks because you are buying a way to see what clients ask AI about your firm, and what AI tells them. Nothing in your reports today shows you that.

A general counsel gets your name from a referral and checks it with AI in private, before anyone calls you. A new CFO at a long-time client asks AI about your firm before approving next year's budget. A board member types "any red flags with this firm" the night before your engagement renews.

Your next client asks AI who to hire. Your current clients ask AI whether to keep you.

What changed

Buyers moved from typing keywords into a search box to asking an assistant a question. In one year, the share of consumers who use Google to research a lawyer fell from 86.7% to 71.9%, while ChatGPT climbed to 41.9%, up from 9% in 2023 (iLawyer Marketing, July 2026). The answer that comes back is a short list of names, and there is no page two. The list is the whole outcome. If you are not on the list, you are not considered.

Does it look at the whole firm only, or at each practice and each lawyer?

AI judges your firm three ways: the firm as a whole, each practice, and each lawyer by name. Most platforms give you one score for the whole firm.

For a firm of 200 lawyers, one number blends your strongest practice and your weakest one into the same result. You cannot tell which is which.

The lawyer level is where it gets personal. AI names individual lawyers as well as firms. A partner who built a practice over twenty years can be missing from every answer about that specialty, while a rival's associate is named twice. Does AI know your best partner exists?

Selectio is the only platform that measures all three layers: the firm, the practice and the lawyer, with continuous monitoring of your firm's progress against what it wants to be known for. Most platforms were built to score a brand with one name, and your clients ask about all three. Read more on how AI builds the shortlist.

Does it ask the questions your clients ask, or a generic list?

Most platforms ask AI the same long list of questions for every client and report how often you show up. Those questions were written for the whole market, and none of them is the one your client asks.

We don't ask AI a thousand generic questions. We ask the ones your clients ask. A thousand generic questions can say you are fine, while the one question your best client asks returns three rivals. The question list has to be built from your firm, in the intake we run with you.

One example. A buyer described a letters-of-credit problem to an AI and never typed the words "law firm." The AI named a New York firm first, because of one bar association article. The buyer never searched for a lawyer. A generic question list would never have found that question, and the firm it named never knew why it was chosen.

Does it show each AI engine separately, or one blended score?

There is no single AI ranking. Each engine trusts different sources, so one blended score hides where you stand on each.

ChatGPT cites Chambers in only 4.1% of legal answers, and Wikipedia in 35.6% (Firesign audit). You can be recommended on one engine and invisible on the next. A blended number averages those two facts into a third that is true of neither.

Your clients use five engines: ChatGPT, Perplexity, Gemini, Claude, and Google's AI Overviews. Some platforms advertise more. Whatever the count, ask to see your results engine by engine.

Does it measure once, or keep measuring?

AI answers change from day to day. A one-time audit is out of date the day you get it.

In a 1,600-answer Perplexity study, about half the firms named showed up in only one of three identical runs (InterCore, 2026). What you see today is a snapshot of one moment.

Continuous measurement means a rhythm. Measure again every month against where you started. Reset the plan every quarter as the engines and your firm both change. We call this cycle the Visibility Loop: fix the record, get competitive, own the space. Ask any vendor what month two looks like. If the answer is "another report," you are buying a document.

Can it show you where AI got its answer, not just a score?

A score alone gives you nothing to do. You need to see which source fed the answer, because the source is what you can change.

In a 1,600-answer Perplexity study, a legal directory rather than a law firm was the first source cited in 77.8% of answers about hiring a lawyer. A third of the firms AI named had no page of their own among the sources. They were described in other people's words, sometimes wrongly (InterCore, 2026).

A general-purpose platform does not know which sources legal answers come from. So it cannot tell you that a directory profile you have not touched in three years is sending clients to someone else. Without the sources, a score tells you something is wrong and nothing about how to fix it.

Does it tell you what to fix first?

Every audit turns up dozens of things to fix. Putting them in order is the whole job. Most platforms just hand you the list.

A real audit finds things like an office listing that no longer exists, a practice page AI cannot read, a partner with nothing online for AI to cite, and a directory profile that sends business to a rival. These are not equal. We rank each one by how much it is likely to move your standing, using fifteen rules of how each AI engine picks legal sources, drawn from published research and tested in our own audits.

That order is the difference between a checklist and a strategy. Small fixes spread everywhere move little. The right fix on the right page moves a lot.

What does it keep after each audit?

Ask what the platform keeps. A record of how AI has treated law firms over time is the one thing a competitor cannot buy later.

Every audit adds to a record of how AI cites law firms, engine by engine, practice by practice. That record turns a score into a trend, and a trend into a forecast.

What if the provider is an agency, not a platform?

The same seven questions apply. One more: ask what you are left with when the engagement ends.

There are good agencies in this market. The difference is what remains when the retainer stops. If the measurement lives with the agency, so does your history. If it is yours, and you can run it again, the position is yours.

---

The market, named

Most platforms in this category are general-purpose software for large brands, and several are very good at that.

The difference that matters is that a brand has one name, and a law firm has a firm, dozens of practices and hundreds of lawyers.

Platforms built for any industry

Each row describes what a product is built for, not how well it performs.

Providers built for law firms*

* Stated models as published, checked September 2026.

---

---