Example data Twelve fictional employees across three example managers. Ratings start as the managers submitted them.
Drag a person to a new box. Or tap a person, then tap Move here. With a keyboard: Tab to a person, then use the arrow keys.
Rating distribution: before and after
Changes
Patterns to check (0)
These are rules of thumb on a small example set, not a statistical test or a verdict. A flag is a reason to look at the evidence again, not proof of bias.
What the flags mean
Heuristics, not verdicts
- Central tendency: 60% or more of ratings at 3, or very little spread. Raters may be avoiding hard calls.
- Leniency or strictness: 70% or more of ratings at the top (4–5) or bottom (1–2), or an average of 4+ or 2 and under.
- Clustering by one manager: one manager’s average is a full point away from everyone else’s, or they gave every person the same rating. It might be a genuinely stronger or weaker team, so ask for evidence.
- Recency hint: the evidence note mentions a strong or weak finish, and the rating follows that recent event rather than the whole period.
- Grid mismatch: the 1–5 rating and the 9-box performance column disagree (1–2 low, 3 moderate, 4–5 high).
The hub’s calibration guide recommends calibrating against scorecard evidence rather than a 9-box. The 9-box is here because many teams still use one, and moving people on both boards shows how two opinions on one grid can disagree.
The thresholds match the rule-of-thumb ratings check on the tools page. Real calibration needs real evidence: goals, feedback and examples from the whole cycle.
Before your real calibration meeting
- How to run a performance calibration meeting
- Calibration meeting template
- How to reduce manager rating bias
- Why some teams are moving away from the 9-box grid
Want a quick read on your own process first? Try the performance review diagnostics.