Most AEO dashboards track vanity metrics that move without revenue moving. Seven metrics actually predict whether AI visibility turns into money: citation rate, share of model, position rank within the answer, sentiment quotient, citation stability, query coverage, and referral conversion rate. The first six measure whether and how you appear in AI answers; the seventh ties it to revenue. Tracking appearance without tracking the conversion behind it is how teams report rising AEO numbers while revenue stays flat. This is what each metric measures, why it predicts revenue, and the one that ties the rest to the bottom line.
AEO measurement has a vanity problem. It is easy to count how often your brand appears in AI answers and report that number going up. It is much harder to know whether that appearance is the kind that drives revenue, because not all citations are equal: a fleeting mention buried at the end of an answer about a query nobody buys from is worth almost nothing, while a stable first-position citation on a high-intent query is worth a great deal. A metric that treats those the same is a metric that misleads.
The seven below were chosen because each one captures a dimension of citation quality that correlates with revenue, not just citation quantity. Track these and your AEO dashboard predicts the bottom line. Track raw mention counts and you get a number that rises while nothing happens.
1. Citation rate
Citation rate is the percentage of your target queries where you appear in the AI answer at all. It is the foundational metric: if you are not cited, nothing else matters. But it is also the one most often mistaken for the whole picture. A high citation rate on low-value queries is a high number that does not pay. Track citation rate against a defined set of queries that actually matter to your business, not against every query you could conceivably appear in, or the metric inflates on traffic that never buys.
2. Share of model
Share of model is your citations as a fraction of all citations on your target queries, which means it measures you against competitors rather than in isolation. This is the AEO equivalent of share of voice, and it predicts revenue better than raw citation rate because it captures the competitive reality: being cited 40% of the time means something very different when your top competitor is cited 80% versus 5%. Share of model replaces rank tracking as the core competitive KPI in AI search, because there is no single ranking to track anymore, only your slice of the answers.
3. Position rank within the answer
Position rank is where in the answer you appear: first mention, middle, or a trailing footnote. This matters because position correlates with both user attention and the weight the answer gives your claim. A first-position citation is the source the answer leans on; a trailing citation is corroboration the user may never read. Two products both "cited" can have very different revenue impact based purely on position, which is why collapsing citation into a binary present/absent loses critical signal. The first mention carries disproportionate value, so a strategy that moves you from trailing to leading on the same queries can raise revenue without raising citation rate at all.
4. Sentiment quotient
Sentiment quotient measures how you are described when cited, not just that you are cited. Being mentioned as "a reliable option" versus "an expensive choice" versus "the leading platform" changes whether the citation drives a purchase. The words the answer attaches to your brand are the words shaping the user's decision. This is a quality dimension that raw counts miss entirely: you can have rising citations that are increasingly negative, a number going up while the actual effect goes down. Tracking the adjectives and framing the engines use is how you catch that early.
5. Citation stability
Citation stability measures how durable your citations are across repeated queries and over time. AI answers are not deterministic; the same query can cite different sources on different runs, and citations decay as content ages and competitors publish. A citation that appears once and vanishes is worth far less than one that holds across runs and weeks, because only the stable citation reliably reaches users. Stability separates a lucky one-time mention from a dependable position you can build revenue forecasts on. Volatile citations are real but unbankable.
6. Query coverage
Query coverage is the breadth of distinct queries on which you appear. Citation rate measures depth on your target set; coverage measures how wide a net you cast across the question space your buyers actually use. Broad coverage means you catch buyers regardless of how they phrase their need, which matters more than ever given that AI engines fan a single user prompt into many sub-queries. A brand cited on one phrasing but absent on ten adjacent phrasings has thin coverage and is losing the buyers who asked differently. Coverage and citation rate together describe both how often and how widely you win.
7. Referral conversion rate: the one that ties it to money
The first six metrics all measure appearance. Referral conversion rate measures outcome: of the visitors who arrive from an AI engine, what fraction convert. This is the metric that ties the other six to revenue, and it is the one most teams cannot measure because AI referral traffic is hard to attribute cleanly in standard analytics. But without it, you are flying blind on whether any of the appearance metrics are the kind that pay. A rising citation rate with a flat or falling referral conversion rate is a warning: you are appearing more but converting the same or worse, which usually means you are winning low-intent queries or being described in ways that attract browsers rather than buyers.
The contrarian conclusion is that six of these seven metrics can all rise while revenue stays flat, and only the seventh tells you whether the AEO program is actually working. Teams build elaborate dashboards around appearance because appearance is easier to measure, then report green numbers to leadership while the bottom line does not move. The discipline is to lead with referral conversion rate and read the appearance metrics as inputs to it: citation rate, share of model, position, sentiment, stability, and coverage are the levers, and referral conversion is the result they are supposed to produce. If the levers move and the result does not, you are optimizing the wrong queries. The full set of signals that feed these metrics shows where each one comes from, and your overall readiness score reflects the appearance side, but revenue lives in the seventh.
Sources
- Princeton, GEO: Generative Engine Optimization: the research framing visibility measurement in generative engines. arxiv.org/abs/2311.09735
- Google Analytics 4, traffic attribution: the basis for measuring referral conversion from AI sources. support.google.com
- Website AI Score, share of model vs rank tracking: why share of model is the core competitive KPI. View article
- Website AI Score, AEO scoring signals: the signals that feed each metric. View article
- Website AI Score, the AI credit score: how the appearance side rolls into a single readiness number. View article
