The employee writes the self-assessment. The manager writes the evaluation. The numbers are assigned. The rating determines compensation, promotion eligibility, and organizational standing for the next twelve months.
All of this is determined by two documents written in December about events that occurred in March. The review does not capture performance. It captures narrative.
⸻
The Recency Effect
The review covers twelve months. The human brain retains about six weeks of detailed work memory.
The employee's self-assessment over-indexes on recent accomplishments, because those are the ones they can describe in detail. The major project they delivered in April is a single bullet point. The minor fix they shipped last week gets two paragraphs because the context is fresh.
The manager follows the same pattern. The employee who performed brilliantly for nine months and struggled for three gets rated on the three. The employee who coasted for nine months and delivered a visible win in November gets rated on the win.
The review does not measure the year. It measures the last quarter, filtered through imperfect recall.
⸻
The Narrative Game
The self-assessment is a marketing document.
Every outcome is framed as intentional. Every struggle is framed as growth. Every failure is framed as a learning experience. The employee who was assigned a project that was destined to fail writes about "navigating ambiguity in a complex stakeholder environment."
The manager reads this narrative and evaluates it not against reality, but against the narratives written by the other direct reports. The rating is comparative. The employee who writes the best narrative receives the highest rating, regardless of whether their actual impact exceeded their peers'.
The review rewards articulation, not contribution.
⸻
The Calibration Fiction
After the reviews are written, the calibration meeting occurs. Managers sit in a room and rank their employees against each other. The distribution must fit a curve.
This means that in a team of ten high performers, three will receive an average rating. Not because their performance was average, but because the curve demands it. The system requires a bell distribution, and reality does not always provide one.
The manager returns from the calibration and delivers the rating. The employee who was "exceeds expectations" is now "meets expectations" because the curve was full. The manager cannot explain this without exposing the calibration process. So they do not explain it.
⸻
The Honest Alternative
If you must operate inside this system, play the long game.
Document your impact in real time. Maintain a running log of outcomes, not activities. When December arrives, your self-assessment is a summary, not a reconstruction. When the narrative game is played, your narrative is grounded in specifics, not adjectives.
The review is a system. Systems can be navigated. But only if you acknowledge what they actually measure.
End.