Understanding results
Filtering
Every metric in Pineprompt is computed over a set of responses, and the filter bar is how you choose that set. The same visibility figure can answer a dozen different questions depending on the slice it is computed over, so the filter bar is the primary way you interrogate your data.
The filter axes
- Date range
- The period the metrics are computed over. Trends compare within this window, and the headline change figures compare it against the preceding period of the same length.
- Platforms
- Narrow to one or more AI platforms, for example to compare how ChatGPT and Gemini treat the same prompts.
- Monitors
- Narrow to the responses of specific monitors, which is how you separate markets when you run the same prompts in several countries.
- Countries
- Narrow to responses collected from specific countries.
- Tags
- Narrow to the prompts carrying specific tags, as described on the tags page.
- Brands and segments
- Scope the brand-level tables and share-of-voice comparison to specific brands, or to a segment, which is a named group of brands.
Axes combine with AND logic: filtering to the "comparison" tag and the ChatGPT platform shows metrics computed over ChatGPT's answers to comparison-tagged prompts only. Within a single axis, selecting several values combines them with OR.
What the filter affects
Everything on the dashboard recomputes under the active filter: the headline metrics and their change figures, the four charts, the brand table, and the source and URL rankings. The latest-responses feed honors every axis except the date range, since its job is always to show the newest answers in the current slice.
Note that scoping to a brand or segment also changes the share-of-voice denominator, which becomes the mentions of the brands in scope. This is what makes a segment useful: it turns share of voice into a comparison within a market segment you define.
Filter state
Your filter selection is remembered per project for the duration of your session, so moving between the overview and the source pages keeps the slice you were studying. Opening another project starts from that project's own filter state rather than carrying the previous one across.
A workflow that works
When a headline number moves, you should narrow before you interpret. Filter to the period of the change, then cycle through the platforms one at a time to find where it originates, then narrow by tag to find which kind of prompt drives it. Once the movement is isolated to one platform and one prompt group, the responses and their sources in that slice usually explain it.