You cannot feel this from inside your own business. Whether ChatGPT names you, whether Google's AI Overview cites you, how you stack up against the shop down the road inside the answer a buyer just read — none of it shows on your dashboard. But it is measurable, directly and repeatably. Here's how the measurement works, what a good result looks like, and where the free part ends.
Most owners try to answer this question by opening ChatGPT and typing their own business name. That feels like a test, but it isn't one — you've told the model who you are, so of course it talks about you. The real question a buyer creates is the opposite: they never mention you. They describe a need — "best commercial sign company near me," "reliable HVAC for a historic home," "who should I use for X in this town" — and the model returns a few names. The only measurement that matters is whether yours is one of them, when nobody prompted it.
A real read has to match how buyers actually ask, across the surfaces that actually answer. That means three things done deliberately.
Real questions, not your name. The scan runs the plain-language questions a customer in your category would type — the need, the location, the qualifiers — not searches engineered to surface you. If you only appear when you're named, you don't appear.
Every surface that recommends. Recommendations get made in more than one place now, and they don't agree. A proper measurement probes the web-grounded engines — Google's AI Overview and Perplexity — and the consumer assistants — ChatGPT, Claude, and Gemini — separately, because you can be named on one and absent on another, and that gap is itself the finding. The assistants shipping inside phones and browsers are becoming their own quiet channel, so they count too.
Named, and cited. There's a difference between the model mentioning you and the model trusting you enough to cite you as a source. Both get recorded, per surface, because they call for different fixes.
Those readings roll up into a single score — an AI Answer Inclusion Rate: across the questions your buyers actually ask, how often are you in the answer, and how strongly? We publish it against the AAIR standard, a public, category-agnostic way to measure the one thing that now decides whether a model puts you in the sentence. It's deliberately not a vanity metric. It's a rate, tied to real buyer questions, that you can watch move over time.
Because the standard is public, it does two useful things at once. It gives you a plain answer to "where do I actually stand" — a 20 means you're absent from four of five answers; an 80 means you're the default the model reaches for. And it gives the market a shared ruler, so a score means the same thing for you as for the competitor you're measured against. Businesses that measure well can carry that as a verifiable signal — the same way a credit rating or a safety certification travels — rather than just asserting they're good at this.
People expect a single verdict. The useful output is a map. A healthy read isn't "we scored high" — it's knowing precisely which questions you win, which surfaces you're strong on, and exactly where a competitor is being named and you aren't. A signage firm might be named confidently by Perplexity for "commercial signs" but invisible in ChatGPT for "monument signs for a subdivision entrance" — two different gaps, two different fixes. The value of the measurement is that it turns a vague anxiety ("are we losing to AI?") into a specific, addressable list.
And the map is honest about the bad news, because the bad news is where the work is. If a competitor is the default answer for your most valuable question, you want that on the table on day one, not discovered a year later when the leads have gone quiet.
Here's the line we draw plainly: measuring is free, and anyone can be measured. The scan, the score, the map of where you're named and where you're absent — that costs you nothing and commits you to nothing. We think the standard should be open, because a ruler only works if everyone can pick it up.
What the number can't do is change itself. Knowing you're invisible to an AI is not the same as becoming visible to it, and the gap between the finding and the fix is exactly where a report dies on a shelf. Closing it means rewriting the pages the model can't read, injecting the structured data that translates your value into something it can parse, and shipping the local-signal changes that put you back in the running — on a site you own, then kept current as the engines keep moving. That's the work. The measurement just tells you, precisely, where to point it.
Can't I just do this myself in ChatGPT?You can start — ask it a buyer's question without naming yourself and see who it returns. What's hard to do by hand is run it across every surface consistently, separate "named" from "cited," and repeat it reliably enough to see the number move. And a one-time check tells you nothing about the direction you're heading, which is the part that matters.
Is the score just marketing, or does it mean something?It's a rate tied to real buyer questions, measured against a public standard, so it means the same thing across businesses — that's what separates it from a vanity metric. It's also a moving number, not a trophy, because your position re-scores as the engines and your competitors change.
What does the free scan actually cost me?Nothing, and it doesn't obligate you to anything. You get the map of where you stand. If you decide to act on it, that's a separate conversation — and if you don't, you still walk away knowing exactly where the AI names you and where it doesn't.