Every category that matures gets a neutral way to keep score — J.D. Power for cars, Nielsen for audiences. AI answers have none. So right now, any firm selling "AI visibility" can also grade its own homework. That's a problem no amount of good intentions fixes. The only fix is a measure that stands on its own.
Ask ten agencies whether you're visible to an AI and you'll get ten answers, each conveniently backed by that agency's own dashboard. One says you're "winning." Another says you're "at risk." A third invented a proprietary index with your logo already in a red zone, and — what luck — a retainer to fix it. None of these numbers can be checked, because none of them are defined anywhere you can read. They're not measurements. They're marketing with a y-axis.
This isn't a knock on the agencies. It's what every young category looks like before a neutral measure shows up. Car quality was once whatever the dealer told you it was. Television audiences were whatever the network claimed. Then J.D. Power and Nielsen put a number on the table that no single seller controlled — defined in the open, repeatable by anyone — and the whole market had to grow up around it. AI visibility is at that exact moment.
A real measure has two properties a private dashboard never will. First, it's published: the method is written down where you can read it, so you know exactly what the number counts. Second, it's reproducible: run the same method again — you, a competitor, a skeptical third party — and you get the same number. If either of those is missing, what you have is an opinion wearing a chart.
The reason this matters so much in AI specifically is that the thing being measured is invisible to the buyer. You cannot feel whether ChatGPT names you. You can't watch Google's AI Overview decide. So when someone hands you a score and says "here's how you're doing," you have no way to push back — unless the method is public and you can run it yourself. Without that, the person grading you is the same person selling you the fix. That's the conflict the whole category is currently built on.
Two things, and they have to be separate. The first is a standard: an open definition of what AI-answer visibility means and exactly how it's scored — which engines, which questions, how an answer counts as naming you, how a score is computed. Written down, versioned, and public, so a number means the same thing no matter who runs it.
The second is an issuer that's independent of any engagement — a body that publishes the standard and stands behind the score without also selling you the work to improve it. That's the J.D. Power part, and the Nielsen part: the measure stands on its own, or it's worthless. The moment the scorer profits from your score, the score stops meaning anything. Independence isn't a nice-to-have bolted on at the end. It's the entire product.
That's why PRAGMA is building the ruler as well as the work. The standard is the AAIR standard — a published, reproducible definition of AI-answer visibility and how it's scored. The independent issuer is the AI Visibility Institute, which publishes that standard and maintains a public registry and seal that anyone can check. And because a score you can verify is more useful than one you're told, the results are published as category rankings — trade by trade, metro by metro — the same way a J.D. Power ranking or a Nielsen rating is public, not private.
We ran it on ourselves first, because a measure you won't apply to your own business is just a sales tool. But the deeper point is structural: the standard and the Institute exist so that "how visible am I to AI" has an answer you don't have to take anyone's word for — including ours. If the category is going to be trusted, someone has to build the neutral yardstick and let it be checked. That's the work, and it's the part that has to stand on its own.
Isn't a standard you built just as biased as anyone else's?It would be, if you had to take our word for it. The difference is that the method is published and reproducible — anyone can re-run it and get the same number — and it's issued through an Institute independent of any engagement. You don't have to trust us. You can check the work.
Why does the issuer need to be independent?Because a measure only means something if the scorer has nothing riding on your score. The moment the same entity grades you and sells you the improvement, the number is compromised. Independence is what makes it a J.D. Power instead of an ad.
Where can I see the rankings?The public registry lists every verified credential, and category rankings — by trade and metro — publish from the same reproducible standard. See the registry here.