Why Metrics Matter Now
AI visibility is now measurable and operationally tracked at thousands of brands. The vocabulary, however, is still consolidating. This reference explains every meaningful AI visibility metric in 2026, what each measures, when it matters, how to interpret a healthy versus unhealthy value, and which metrics to prioritise at each stage of an AI-visibility programme.
The Core Six Metrics
The metric vocabulary has converged on six core measures. Every credible AI visibility platform reports these or close equivalents.
1. Mention Rate
The percentage of monitored prompts where the brand is named at all in the AI response. The most binary signal, useful as a baseline. A healthy mention rate for established brands sits between 35 and 70 percent depending on category competitiveness; below 15 percent indicates structural absence; above 80 percent suggests the prompt set under-samples competitive queries.
2. AI Share of Voice
The brand's mention rate divided by the sum of mention rates across the brand and the tracked competitor set. A direct analogue of share-of-voice from traditional brand tracking, rebased on AI responses. Healthy share-of-voice values vary sharply by category structure; in concentrated categories (3 to 5 dominant players) the leader typically holds 35 to 50 percent; in fragmented categories (10+ players) the leader may hold only 18 to 25 percent.
3. Citation Rate
The percentage of citation-eligible responses that cite at least one URL from the brand's domain. Distinct from mention rate, citation rate measures whether the brand is the source AI grounds its answer on, not just whether the brand is named. Citation rate is exposed cleanly on Perplexity, ChatGPT Search, Google AI Overviews, and Copilot Web; default Claude and default ChatGPT do not expose citations and require paraphrase inference.
4. Citation Position
For responses where the brand is cited, the average ordinal position of the citation. First-position citations earn 4 to 5x the click-through of fifth-position citations. Tracking position alongside citation rate prevents the common mistake of celebrating citation gains that occur in low-CTR positions.
5. Framing Score
A composite score of how the AI describes the brand when it does mention it. Adjective tone, comparison ordering, and qualifier presence (positive, negative, hedged) all feed the score. A brand can have rising mention rate while framing score declines, often a signal that competitors are expanding share through superior earned authority while the focal brand expands through volume alone.
6. Hedge Tier
For responses where the brand is recommended, the strength of recommendation (singular winner, ranked shortlist, unranked shortlist, mention-only). Hedge tier varies sharply by AI platform, Claude hedges most heavily, ChatGPT default mode hedges least. Tracking hedge tier reveals whether the brand is converting awareness into recommendation power, which is what actually moves buying decisions.
Three Auxiliary Metrics
Beyond the core six, three auxiliary metrics matter in mature programmes.
Sentiment Trajectory
The 30-day rolling sentiment of AI brand mentions. Sentiment can decay even when mention rate holds steady, often the leading indicator of brand reputation issues that have not yet surfaced in traditional channels.
Source Diversity
The number of distinct domains AI cites when grounding mentions of the brand. Low source diversity (heavy reliance on a single source, such as Wikipedia) is fragile because a single edit can cascade into AI visibility loss. High source diversity is structurally more durable.
Recovery Time
The median time between a citation loss and a citation recovery (when the same query is monitored continuously). Recovery time tells you how operationally responsive the AI visibility programme is. Mature programmes drive recovery time below 14 days for monitored queries.
Metric Priority by Stage
| Programme Stage | Top Priority Metric | Secondary Metric |
|---|---|---|
| Baseline / discovery | Mention rate | AI share of voice |
| Optimisation | Citation rate | Citation position |
| Maturity | Framing score | Hedge tier |
| Crisis / regression | Sentiment trajectory | Recovery time |
Common Measurement Mistakes
Three mistakes show up in 80 percent of AI visibility programmes. First, optimising mention rate while citation rate stagnates, this often indicates volume content gains that do not translate into the source-authority signals AI grounds on. Second, declaring victory on first-page citations while average citation position drifts toward five and below, the visibility looks fine on paper but converts poorly. Third, tracking only English-language responses for global brands, which systematically misses regression in non-English markets where competitors are expanding.
How Presenc AI Helps
Presenc AI reports all six core metrics plus the three auxiliary metrics across every major AI platform in one dashboard. Each metric is tracked per platform, per query, and per competitor, so the diagnostic question is never "what is happening" but "where exactly should we act first". For brands moving past baseline measurement into operational AI visibility, the metric layer is the foundation.