People Day
The People Day report page presents the visuals listed below. Each row gives the on-screen name, its underlying metric, and a short description; see the metric glossary for the full formula and interpretation.
Visuals:
| Name | Measure | Description |
|---|---|---|
| Rating changed by moderator | Count of reviews where Moderated = Yes÷ count of all reviews | Share of reviews where the moderator (the calibrating layer, typically the manager's manager) changed the rating during moderation, rather than leaving the manager's rating as is. It signals how much calibration is actually altering outcomes: a high rate means moderation frequently overrides, a low rate means manager ratings mostly stand. Because it's a share of the total in view, filtering re-bases it. It counts reviews, not people. |
| Worker/manager alignment | Count of reviews, by initial alignment (Yes / No / Not Available) | How many reviews had the worker's self-rating agree with the manager's rating at the initial stage, split into Yes (they matched), No (they differed), and Not Available (no worker rating to compare). It gauges how aligned self-assessment and manager assessment are before any final calibration; a large No slice signals gaps between how workers and managers see performance. It counts reviews, not people. |
| Manager review change (people day) | Count of reviews, grouped by the manager's submitted rating, then by the published (post-moderation) rating | How manager ratings shift from submission to publication, shown as a two-level ring: the inner level is the rating the manager originally submitted, the outer level is the rating actually published after moderation, and the size is the number of reviews following each path. Where the outer rating differs from the inner one, the rating was changed during moderation, so the visual surfaces both how much changed and in which direction. It counts reviews, not people. |