Definition
AI visibility is the degree to which a brand is recognised, retrieved, recommended and cited by AI systems when users ask questions in its category. It is to assistants what search visibility was to search engines — except it is measured in answers, not rankings.
A brand's AI visibility can be decomposed into a funnel of five measurable stages. Each stage gates the next, and each leaks for different reasons — which is why "how visible are we?" is less useful than "at which stage do we disappear?".
| Funnel stage | Question it answers | Typical failure cause |
|---|---|---|
| Recognised | Does the model resolve your brand as a distinct entity? | Ambiguous naming, inconsistent identity data |
| Retrieved | Does your evidence appear in relevant answers at all? | Thin or uncrawlable canonical content |
| Recommended | Are you named in recommendation contexts? | Weak independent consensus |
| Preferred | Are you the first-mentioned option? | Undifferentiated positioning |
| Trusted | Are your own pages cited as sources? | Nothing verifiable or data-led to cite |
The AI Visibility Funnel. Diagnosing the deepest leaking stage tells you which intervention to fund.
AI Search vs Traditional Search
Traditional search visibility was positional: ten links, a rank, a click-through curve. AI visibility is compositional: the assistant writes one answer, chooses whom to name, and decides how to frame each name. Three consequences follow.
- Concentration: a typical commercial answer names two to five brands; the top three capture more than half of the recommendation weight in most categories.
- Framing: sentiment and justification are part of the result — visibility includes how you are described, not just whether.
- Volatility: model updates reshuffle answers the way algorithm updates reshuffled rankings, but faster and with less announcement.
Because assistants disagree with each other on roughly half of advisory prompts, aggregate "AI visibility" numbers hide model-level risk. A brand can dominate one assistant and be invisible in another while its dashboard shows a healthy average.
How AI Visibility is Measured
Credible measurement uses a fixed panel of real buyer prompts, issued repeatedly to each model family, with responses parsed into entities, recommendations and citations. From those primitives, a composite score can be computed — provided its construction is published.
The visual communicates that a visibility score is an aggregation of independently reproducible measurements, not a single opaque number.
Any composite score embeds editorial weighting choices. Trust scores whose weights, panels and limitations are published; treat hidden-formula scores as marketing.