Productivity
Run a talent calibration
A nine-box built from ratings, with the biases named.
Fill it in
Performance and potential across the org.
Recency, tenure and manager leniency show up in the distribution.
A grid with no actions is a picture.
Your prompt
Build the calibration. Name the biases visible in the DISTRIBUTION before discussing anyone. Recency, a manager who rates everyone highly, and tenure standing in for potential all show up as shapes: one department with no low ratings, potential correlating with years served. Calibration that skips this calibrates the biases. Plot the grid and show the distribution per manager alongside it, so an outlier rater is visible. Say what each box means as an ACTION. A nine-box with no actions is a picture that makes people uncomfortable and changes nothing. Flag anyone whose rating moved a long way without an event. Use no names in the summary view.
Use Run a talent calibrationOpens with everything above already filled in.
Why this works
Calibration that skips the distribution calibrates the biases. Recency, a lenient manager and tenure standing in for potential all show up as shapes before anyone is discussed. This names them first, then plots the grid, with an action per box.