Define the measurement decision
Create distinct Prompt Groups with the same market and platform scope, then compare patterns without merging their denominators.
Measurement method
| Step | Action |
|---|---|
| 1 | Define a category group without the brand or product name |
| 2 | Define product and feature groups using the language buyers actually use |
| 3 | Keep comparison and problem-solving prompts separate from navigational checks |
| 4 | Report coverage, position, Sentiment, and citations within each group |
| 5 | Escalate discrepancies: category absence needs discovery work; product inaccuracy needs fact correction |
Worked calculation or observation
A company can be frequently described when its product is named yet absent from 'best tools for X' prompts. One combined visibility percentage would make the product look healthy while hiding weak category discovery.
Evidence, limits, and UnderAI's role in broad-topic and product tracking in AI search
Keep these fields with the decision:
- metric name
- numerator
- denominator
- scope
- platform
- Prompt Group
- success state
- observation date
- formula version
Group-level scores are scoped measurements, not market share. Results depend on the chosen questions, platforms, market, and date.
UnderAI GEO Workspace supports project-defined Prompt Groups and filters, making the category-versus-product split explicit in daily observations.