Define the measurement decision
Choose the definition that matches the decision, then keep the denominator visible in every report.
Measurement method
| Step | Action |
|---|---|
| 1 | Define the eligible answer set and missing-data rule |
| 2 | List the brands and aliases included in the comparison |
| 3 | Decide whether an answer contributes mentions, positions, or citation presence |
| 4 | Calculate results by Prompt Group and platform before aggregating |
| 5 | Review the answer text so a mention in a warning is not treated like a recommendation |
Worked calculation or observation
For a 30-prompt comparison group, mention share and top-three recommendation share can tell different stories. Reporting both is clearer than collapsing them into a proprietary score.
Evidence, limits, and UnderAI's role in AI search share of voice
Keep these fields with the decision:
- metric name
- numerator
- denominator
- scope
- platform
- Prompt Group
- success state
- observation date
- formula version
Prompt design and competitor selection strongly affect the result. Changing either breaks comparability with the previous series.
UnderAI provides Coverage, Visibility, TOP3, competitor observations, Sentiment, and snapshots; teams should name the exact metric instead of relabeling all of them as share of voice.