What it is

AI Visibility metrics measure how often, how prominently, and how positively your brand appears in AI-generated search results.

These metrics power:

The Overview page
The Search terms table
Detailed term views
Competitor comparisons
Reports and exports

This article explains the formulas and aggregation logic behind each core metric.

Why it matters

AI search is probabilistic.

Unlike traditional SERPs:

Results vary per run
Mentions are contextual
Ranking is narrative-based
Citations are model-dependent

Without understanding the math behind these metrics, it’s easy to misinterpret performance.

These formulas ensure:

Frequency matters
Placement matters
Consistency matters
Volatility is visible

How AI Visibility data is generated

For each search term:

The prompt is executed against the selected AI engine.
The response is analyzed.
Brands are detected.
Positions, mentions, sentiment, and citations are extracted.
Results are aggregated across runs.

Metrics are calculated across a selected time window:

Latest run
24-hour trailing average
7-day trailing average
30-day trailing average

Core metrics & formulas

Visibility score

Measures how prominently and consistently your brand appears in AI-generated results.

Formula:

Visibility Score = Detection Rate × Rank Score × 100(rounded to 1 decimal place)

It combines:

How often your brand appears
How high it ranks when it appears

Detection rate (0–1 scale)

Detection rate reflects consistency of presence.

Detection Rate = Number of runs where brand was found / Total runs

Example:

If your brand appears in 80 out of 100 runs

Then Detection Rate = 0.8 (80%)

Rank score

Rank Score rewards higher placements.

Rank Score = 1 / (1 + 0.1 × (Average Rank − 1))

Where:

Average Rank = average position across runs where the brand appeared
Position 1 = first mentioned

Lower average rank → higher rank score.

This formula ensures:

Position 1 performs significantly better than Position 5
Improvements near the top have stronger impact
Deep placements contribute less to visibility

Position

Indicates where your brand appears in the AI-generated response.

Position 1 = mentioned first
Lower number = stronger placement

Depending on the selected timeframe, Position may represent:

Latest position (most recent run), or
Average position across selected period

⚠️ Important:
Position is calculated only for runs where your brand was detected.

Top 3 rate

Measures how often your brand appears in premium placements.

Top 3 Rate = (Runs where Position ≤ 3 ÷ Total runs) × 100

This reflects consistency of high placement, not just presence.

Mentions

Total number of times your brand is detected across runs.

Important distinction:

Detection rate counts runs
Mentions counts occurrences

If your brand appears twice in one response, both mentions are counted.

Mentions measure narrative depth, not ranking.

Citations

Counts how many times your brand is referenced alongside supporting links or attributed sources within AI-generated responses.

Citations are extracted from:

Linked domains
Attributed references
Structured source mentions

For full attribution logic, see:
Citations, sources & attribution in AI results

Sentiment

Indicates how positively, neutrally, or negatively your brand is described in AI responses.

Sentiment is calculated by:

1. Classifying contextual language surrounding brand mentions
2. Aggregating sentiment across runs

Sentiment score reflects:

Percentage of positive mentions out of total sentiment-classified mentions.

It does not directly affect Visibility Score, but provides perception insight.

Time windows & aggregation logic

Metrics can be viewed as:

Latest run
24-hour trailing average
7-day trailing average
30-day trailing average

Trailing averages smooth volatility caused by:

Model randomness
Prompt variation
Data refresh timing

This helps distinguish short-term fluctuation vs sustained performance changes.

Edge cases & interpretation notes

Edge Cse	Interpretation
Zero detection	If your brand is not detected in any runs: Detection Rate = 0 Visibility Score = 0 Position and Rank Score are not calculated.
High detection, low visibility	If Detection Rate is high but Visibility is moderate: Your brand appears often, but at lower positions.
Low detection, high rank	If Detection Rate is low but Rank is strong: You rank well when you appear, but lack consistency.
Volatility	AI engines may produce different outputs across runs. This is expected behavior. Use 7-day or 30-day averages for strategic decisions.

How these metrics work together

Metric	Measures	Strategic meaning
Detection rate	Frequency	Are you consistently included?
Position	Placement	How early are you mentioned?
Visibility score	Combined strength	Overall AI presence power
Top 3 rate	Premium placement	Are you top-tier?
Mentions	Narrative depth	How much are you discussed?
Citations	Authority signals	Are you source-backed?
Sentiment	Perception	How are you described?