Why Performance Reviews Feel Arbitrary

Reviews are widely regarded as unsatisfying by the people receiving them and the people conducting them. The reasons lie in the structure rather than in the individuals involved.

One conversation is asked to do several jobs

A review is expected to develop the person, record a judgement for pay purposes, compare them with colleagues and document performance for future decisions.

These aims conflict. Honest discussion of weaknesses is useful for development and costly for someone whose pay depends on the same conversation, so candour is suppressed on both sides.

What emerges is a hybrid that serves none of the purposes well. Development is diluted, and assessment is softened.

Judgement is compressed into a period nobody remembers

A year of work is assessed from memory, and memory favours the recent and the vivid. Steady contribution across months is systematically underweighted.

Any conspicuous event near the review date carries disproportionate weight, whether or not it is representative. The timing of incidents affects outcomes more than people expect.

Contemporaneous notes reduce this substantially, which is why the practice is so often recommended and so rarely maintained.

Ratings are constrained before individuals are considered

Many systems require the distribution of ratings to fit an expected shape. Where that applies, an individual assessment is partly a decision about the group.

Someone can perform well and receive a middling rating because the available positions were used. The rating describes a position in a queue rather than a standard that was met.

Since the constraint is usually not stated plainly, the outcome appears personal when it was largely arithmetic.

Attribution is genuinely hard

Most work is joint, and separating one person's contribution from a team's result is difficult even in good faith. Visibility becomes a proxy for contribution because contribution is not directly observable.

People whose work is coordination or prevention suffer most. Absence of problems is real value that leaves no evidence.

Making such work legible during the year, rather than arguing for it afterwards, is the more effective approach. The assessment can only weigh what it can see.

The useful parts can be separated out

Feedback that changes behaviour works best close to the event and detached from any pay decision. Nothing about that requires an annual cycle.

Treating the formal review as a record rather than as the occasion for learning removes much of the disappointment. The learning happens elsewhere, more often, and with less at stake.