Definition
Performance calibration is the process managers and HR use to compare and adjust performance ratings across teams before reviews are finalized, so the same rating means the same thing regardless of who gave it.
PeopleManagingPeople's guide to the practice describes it as managers and HR leaders reviewing and adjusting ratings to ensure consistency and fairness across the organization. Deel's guide to calibration meetings frames the meeting itself as a structured session to evaluate and align ratings across evaluators.
The problem calibration fixes
Two employees can do equally strong work and land on different ratings, not because their work differed, but because one manager rates generously and another rates strictly. Left alone, that gap compounds: it skews who gets a bonus, who gets promoted, and who ends up on a performance plan, based on which manager they happen to report to rather than what they actually delivered.
Calibration exists to catch that gap before ratings become final, not after. Once a rating is communicated to an employee and tied to pay, walking it back does more damage than the inconsistency it was meant to fix.
How a calibration session runs
Step — What happens · Managers submit draft ratings
- Each manager rates their own team first, independently — Ratings are compared across teams · HR or a calibration lead looks for patterns: one team rated unusually high or low relative to others
- Managers defend outlier ratings with evidence — A manager explains why their team is genuinely stronger this cycle, not just rated generously · Ratings are adjusted where evidence doesn't hold
- Changes get made before ratings are communicated, not after — The standard gets documented · What "meets expectations" looked like this cycle, for next time
The session works only when managers bring evidence, not opinions. "My team worked hard" doesn't survive calibration. "Here's what each person delivered against their goals" does.
What good calibration evidence looks like
A rating defended with a specific example, tied to a goal or a competency, holds up. A rating defended with "they're one of my best people" does not, regardless of how true it might be. The distinction matters because calibration sessions run on limited time, and unevidenced ratings are the ones that get flagged, adjusted, or argued over longest.
How Spark.work supports the calibration conversation
Spark.work doesn't run calibration meetings itself, but it holds the evidence a calibration session needs. Employee Reviews keeps rating history and manager comments in one place instead of scattered spreadsheets, so a calibration lead can actually compare ratings across teams rather than requesting exports first. 360 Performance Reviews adds peer and self-assessment alongside the manager rating, which gives a calibration discussion more than one manager's opinion to weigh.
What Performance Calibration Is Not
Often confused with — The difference
- A performance review — A review is the assessment of one person. Calibration compares ratings across many people and teams before those individual reviews are finalized.
- A forced ranking — Forced ranking requires a fixed distribution, such as a set percentage rated top or bottom regardless of actual performance. Calibration adjusts for inconsistency in how ratings were given; it does not require a predetermined shape.
- A pay decision — Calibration adjusts the rating. What that rating then means for pay or promotion is a separate decision that happens after calibration, not during it.