How to Run a Performance Review in a Small Team

What a review is for

A review is not for delivering feedback. Feedback belongs in the week the thing happened, which is the argument in our guide to giving feedback that doesn't damage trust.

A review does three things that weekly feedback cannot. It looks at a whole period and asks whether the trajectory is right. It creates a written record, which matters both for fairness and for any difficult conversation later. And it makes space for a discussion about where this person is going, which never fits into a normal week.

Preparing properly

1. Collect examples across the whole period, not the last month

   Recency bias is the biggest single distorter of reviews. Something that went wrong three weeks ago will outweigh six good months unless you deliberately correct for it.

   The defence is notes taken as things happen. If you keep a written record from your 1:1s, this preparation takes twenty minutes instead of two hours of unreliable recall.

   The most important step

2. Ask two or three colleagues for input

   One thing that works well, one thing that could be better. Short, informal, and told to the person it concerns in general terms rather than attributed.

   Colleagues see things managers do not, particularly how someone is to work with rather than what they produce.

   Keep it light

3. Ask for a short self-assessment in advance

   What went well, what was hard, what they want next. Half a page.

   The gap between their view and yours is usually the most useful thing in the conversation, and it is far better discovered before the meeting than during it.

   Half a page, sent ahead

4. Write your view down before the meeting

   Not to read aloud, but because writing forces you to check whether your assessment is supported by examples or by an impression.

   This is where you notice that a general sense of someone being difficult rests on one incident in March.

   Tests your own bias

5. Decide the pay question separately, beforehand

   Know your answer on compensation before you walk in, and do not let the conversation change it in the room.

   Deciding pay live rewards whoever argues best on the day, which is not the same as whoever contributed most.

   Decide before, communicate after

The conversation

Sixty minutes, private, with no laptop between you.

Their view first. Walk through their self-assessment. Ask questions rather than correcting. Going first means they set out their case before hearing yours, which produces a more honest version of it.

Your view second, with examples. Strengths with specific instances, then the areas to develop with specific instances. General praise and general criticism are equally useless: neither can be acted on.

Then the trajectory conversation. Where do they want to be in a year, what would need to be true, and what would they need from you. This is the part that most affects whether good people stay, and it is the part most often cut when the meeting overruns.

End with two or three things for the next period. Written down, specific, and revisited in 1:1s rather than saved for the next review.

Send a short written summary within a day. Not a form, just what you discussed and agreed. Both of you will remember it differently in four months, and the written version prevents that becoming a problem.

Keep pay separate

Hold the compensation conversation in a different meeting, ideally a week or two later.

When pay is on the table, people cannot hear developmental feedback. They are waiting for the number, and everything before it is processed as either justification or padding. Separating them is the single easiest improvement most small companies can make to their review process.

If a raise is not happening, say so plainly and explain what would change that. Vagueness here produces someone who spends six months guessing and then leaves.

Common mistakes

Rating everyone as meets expectations to avoid difficulty. It tells strong performers their work is unremarkable and struggling people that everything is fine. Nobody is served.

Saving up criticism. Six months of withheld feedback delivered in one hour is not a review, it is an ambush, and it damages trust in a way that takes a long time to repair.

Reviewing personality instead of work. "Needs to be more confident" is not actionable and often smuggles in bias. Describe behaviour and effect, not disposition.

Comparing people to each other in the room. Ranking is sometimes necessary internally. Saying it aloud in a review poisons the working relationship between colleagues.

Skipping it when things are busy. The skipped review is usually the one with the struggling person, because it is the uncomfortable one. Our guide to handling an underperforming team member covers what happens when that conversation is deferred: the situation hardens from solvable into personal.

Reviewing without the weekly foundation. A review works because it summarises a period in which feedback was already flowing. Without regular 1:1s, described in how to run 1:1s that produce decisions, the review has to do all the work at once, and it cannot.