Why annual appraisals often arrive too late

Appraisal systems are usually asked to do too much at once: rate performance, plan development, signal pay, gather evidence and repair a relationship, all inside one uncomfortable hour that both people have been quietly dreading since the reminder email arrived.

The design flaw is not the form, although the form rarely helps. It is the timing.

Feedback has a half-life

Behaviour changes most easily close to the moment it occurred, while the detail is still shared and the emotional temperature is low enough to think. Wait nine months and you are no longer discussing a specific event. You are discussing two competing recollections of it.

This is why so many appraisal conversations can feel oddly abstract. "Sometimes you can come across as quite directive in meetings" is not feedback. It is a summary of feedback that was never given. The person on the receiving end cannot picture the moment, cannot test the claim, and cannot easily do anything with it except feel vaguely accused.

What the annual cycle quietly rewards

An annual process tells managers, in effect, that performance conversations are an event rather than a habit. That is a comfortable message for anyone who finds those conversations difficult, which is most people.

It also creates the recency problem, where the last two months dominate the rating for twelve. And it produces the familiar surprise: a person who believed things were broadly fine discovering in March that they have not been fine since the previous autumn. That surprise does more damage to trust than the content of the feedback ever does.

There is a further, subtler cost. Dependable people receive the least attention, because there is nothing urgent to write. Being reliable becomes a route to being overlooked.

Appraisal as documentation

Where the process is largely defensive — evidence for pay decisions, for capability procedures, for the auditor — the incentives shift again. Managers write to protect themselves. Ratings cluster in the safe middle. Calibration meetings adjust numbers for distribution rather than accuracy. Everyone completes the process on time and nobody's behaviour changes.

That is not a failure of effort. It is what the system was, in practice, designed to produce.

What tends to work better

Keep the annual conversation, but demote it.

Its job is to look back at the year as a whole, agree development priorities and hold an honest conversation about direction. It should contain no surprises, because everything in it has already been said.

Move the real work into short, frequent, specific conversations — a fortnightly or monthly check-in of twenty minutes where the manager notices something concrete and says so. Praise and correction both belong here. The skill is specificity: what happened, what its effect was, and what would be more useful next time.

Separate the pay conversation from the development conversation. People cannot hear coaching while they are waiting to find out about money.

And give managers practice, not just training. Most know what good feedback looks like. What they lack is confidence in the moment when the other person's face changes. I like to call this ‘real playing’ (see what I did there!), practicing real conversations so the uncomfortable becomes less so.

A modest test

Ask a manager to describe, without notes, the two things each of their people is working on developing. If they can, the conversations are happening. If they reach for the appraisal file, the ritual has replaced the mechanism.

“By the time the form is filled in, the moment for the conversation has usually passed.”