How to run a performance review that helps both sides

A performance review is a structured conversation that answers three questions with evidence: how did the period actually go, what happens next (goals, development, sometimes pay), and what does each side need from the other? Reviews earn their bad reputation when they are improvised monologues delivered from memory; run from records, with structure, they become the most useful hour a manager and employee spend all year.

Preparation: the review is won before the meeting

• Pull the record, not the memory — the goals set last period, completed training, recorded achievements, incidents, and the previous review’s commitments. Memory-based reviews systematically reward the last six weeks and the loudest personalities.
• Ask for a self-assessment first — the same three questions you will answer: what went well, what fell short, what should change. The gap between self-assessment and manager assessment is the meeting’s real agenda.
• Write your assessment before the meeting — a page: outcomes vs goals, strengths with examples, gaps with examples, proposed next goals. Writing first prevents the meeting from negotiating your conclusions into vagueness.
• No surprises rule — anything significant in the review must have been raised when it happened. If it was not, that is feedback for the manager, and it waits for the next occurrence rather than ambushing this meeting.

The meeting itself

Structure beats improvisation: start with their self-assessment (you learn more listening first), then your assessment — specific, example-based, praise and criticism both tied to recorded events — then converge on next-period goals and any development plan, and end with the reverse question: "what should I do differently as your manager?" asked like you mean it. Keep pay mechanics out of the development conversation where your process allows; the salary number otherwise deafens everything said before it. Sixty minutes, no phones, never cancelled — a rescheduled review teaches everyone the real priority of performance.

Ratings, calibration, and fairness

If you use ratings, define the scale in behaviors (what does "exceeds" actually look like for this role?) and calibrate across managers — the same performance must earn the same rating in different teams, which requires managers comparing notes with evidence. Uncalibrated ratings are how identical work earns different raises, and employees discover it faster than management assumes. If you cannot invest in calibration, prefer written assessments without numeric scores; a defensible paragraph beats an indefensible number.

Afterward: the part that makes it real

Document the outcome — assessment summary, agreed goals, development commitments — dated and visible to both parties, and record it where it persists: in EmployDB, reviews land in the employee’s record alongside the training and achievements they discussed, so next period’s review starts from this one instead of from zero, and a manager change does not erase the history. Then honor the commitments between reviews: the development plan that is never mentioned again until next year is the single most common way review processes lose credibility.