The AI visibility metrics that actually matter

If GEO is the work, metrics are how you know it is working. The problem is that "are we visible in AI?" is not one number — it is a small stack of them, each answering a different question. Here are the five that matter and how to read each.

A single blended score (typically 0–100) that rolls up how often you are mentioned, how prominently, and how positively, across every engine and every tracked question. It is your headline number — useful for tracking direction over time, but only meaningful when you can drill into what moved it.

Of all the brand mentions across your tracked answers, what percentage are you — versus each competitor? Share of voice is zero-sum and therefore honest: it shows exactly who is winning the conversation and by how much. A rising visibility score means little if a competitor's share is rising faster.

Being named is not the same as being named well. Sentiment captures whether each engine talks about you as a leader or an also-ran — positive, neutral or negative — per engine. A pile of neutral or hedged mentions is a different problem than being ignored, and it needs a different fix.

When you are named alongside others, where do you land — first, or buried at the end of the list? Order inside an answer carries real weight; the first brand mentioned gets disproportionate attention. Average position tells you whether you are leading the sentence or trailing it.

Which sources does each engine cite to justify its answer, and how often is one of them yours? Low citation share with high mentions means the engines talk about you using other people's pages — a signal to invest in your own answer-first content and in earning citations on the sources that already rank.