The phrase "AI visibility" got thrown around so much last year that it stopped meaning anything. Agencies slap a score on a slide, claim a 40-point jump this quarter, and everyone nods politely. Most of those scores are nonsense, built on a hand-picked set of five questions that happen to flatter the agency.
I want to give you a definition that survives contact with a client. AI visibility is not one number. It is the measurable degree to which an AI engine surfaces, cites, and describes your brand when a buyer asks about your category, your city, or your specific problem.
Why the old metrics stopped telling the truth
For fifteen years, SEO ran on positions. You ranked, you waited for the click, you counted the sessions. That model broke the moment an AI summarized the answer and the user never scrolled.
The awkward reality most people miss: a brand can now lose the answer while still winning the rank. You sit at position two, the AI tool pulls its summary from a competitor's cleaner page, and your organic sessions barely move. Position tracking gives you a false sense of safety. AI visibility tracking exists to catch what the ranks stopped saying.
What we actually measure
We do not track "mentions" in some mystical AI score. We track three concrete things against a fixed, seeded set of queries — usually forty to sixty real questions your buyers type, pulled from GSC and LLM research.
Citation rate. How often your domain shows up as an explicit source, whether linked or named. This is the closest thing to an AI ranking you'll get. It tells you whether the AI trusts you enough to point to you.
Entity presence. How often your brand is named or described at all. You can be cited as a generic source without ever being the named answer. We separate "quoted as proof" from "named as the answer" because they require completely different fixes.
Sentiment and completeness. When AI describes you, does it describe you accurately and favorably, or does it carry a half-remembered, out-of-date paragraph? This surprises people the most: a lot of AI answers about a business are flat-out wrong, and nobody was checking.
Beware the vanity number
A single weighted index is useful for a headline and dangerous for a decision. Different AI engines favor different things, and a ChatGPT-strong brand can be Perplexity-invisible. We report each engine separately, and we weight the query set toward commercial intent — the questions that have a dollar attached — rather than a generic basket of brand terms.
If someone quotes you a one-number AI visibility score built on five queries, ask to see the query set. If they can't show you the questions, the score is a marketing artifact.
The honest version
AI visibility is a leading indicator you can action. Cite one way and the money flows differently than if you're just being described. Start by measuring the three buckets above, fix the attribution gaps, and re-run the same query set on a schedule. The number that matters is the one that changes when you change your content, your schema, and your entity cohesion — not the one that changes when someone tweaks a dashboard's weighting.
At Launch at Dawn we run this exact audit for brands in Montreal and Vancouver, and we're happy to walk yours through it. Book a free website teardown and we'll show you where AI answers are already talking about you — and where they're ignoring you.