Understand your results

Report metrics

This guide explains how Geolity converts real AI responses into the scores, rates, rankings, and evidence displayed in your report.

The evaluation unit

Every buyer prompt is sent to each enabled AI engine. Geolity evaluates the target brand in every prompt × engine answer using three visibility states.

Mentioned

Weight 0.5

The brand appears in the answer or a comparison table but is not recommended.

Not Visible

Weight 0

The brand does not appear in the analyzed answer.

Across engines: the question-level result uses the best observed state — Recommended outranks Mentioned, which outranks Not Visible. Engine-specific views evaluate each engine separately.

Core report metrics

Visibility Score

(Recommended + Mentioned × 0.5) ÷ Total prompts × 100

Rewards a direct recommendation more than a simple mention. The score is rounded to a whole number from 0 to 100.

AI Visibility Rate

(Recommended + Mentioned) ÷ Total prompts × 100

Measures how often the brand is visible at all. Recommendations and mentions count equally for this rate.

Category Rank

Brands ordered by Visibility Rate, then Recommended count

Compares the analyzed brand with competitors automatically detected in the same AI answers.

Example

Across 40 prompts, suppose the brand is Recommended in 12 and Mentioned in 8:

Visibility Score: (12 + 8 × 0.5) ÷ 40 × 100 = 40
AI Visibility Rate: (12 + 8) ÷ 40 × 100 = 50%

How breakdowns are calculated

By scenario

Visible prompts in a buying scenario ÷ total prompts assigned to that scenario. The report also shows the category leader for comparison.

By AI engine

Recommended and Mentioned counts are recalculated using only answers from that engine, with the prompt count as the denominator.

By product

Distinct prompts that recommend or mention a detected product are counted, then scored with the same 1.0 / 0.5 weighting.

Total real queries

Prompt count × enabled AI engine count. Overall question-level metrics still use the prompt count as their denominator after cross-engine best-result selection.

Evidence, answers, and citations

Scores are summaries, but the report keeps the underlying evidence auditable. The evidence section shows the buyer question, scenario, per-engine state, raw answer, and cited sources.

Evidence questions are selected consistently from the run, prioritizing questions where the brand appears while balancing topic coverage.
A citation is displayed as evidence for the answer that used it; it does not automatically increase the Visibility Score.
Using the same prompt bank in future reports makes changes in score, rank, answers, and sources comparable over time.

Ready to apply these metrics? Continue with the report walkthrough.