Leadership · September 18, 2026 · 7 min read
A performance review process that survives a company of 50
Most small companies either skip performance reviews entirely or copy an enterprise process that collapses under its own weight. There is a middle version that takes managers two hours a quarter and produces documentation you can actually rely on.
Why annual reviews fail in growing companies
An annual review is a summary of a year that nobody was documenting. When a manager sits down in December to describe performance, they are reconstructing eleven months from memory, and memory in a busy company defaults to the last six weeks.
The result is a conversation that feels unfair to the employee and produces no usable record for the company. Then the following December the same thing happens, and the pattern itself becomes the problem.
The quarterly rhythm
Four short cycles beat one long one. Each quarter, every manager writes three things per direct report: what was expected, what happened, and what changes next. Two or three sentences each is enough.
The review conversation takes thirty minutes and its purpose is alignment, not scoring. Once a year, the four quarters are compiled into a formal review, which by then is mostly a matter of reading rather than remembering.
This is the single highest leverage change available to a company with under two hundred people, and it costs each manager roughly two hours a quarter.
- ·Quarterly: three notes per person, thirty-minute conversation, no ratings required
- ·Annually: compile the four quarters, discuss development and compensation separately
- ·Immediately: document any serious performance concern within 48 hours, not at cycle end
Write expectations before the work, not after
The most common documentation failure is that expectations were never written down, so a later assessment becomes a matter of opinion. Three or four outcome statements per role, agreed at the start of the quarter, remove most of that ambiguity.
Outcome statements should describe results rather than activity. Owner of the onboarding programme reaching first-week completion above target is a result. Attends meetings and is responsive is not.
Ratings: use them sparingly, define them precisely
If you use ratings at all, define each level in behavioural terms and calibrate them across managers. An uncalibrated five-point scale quietly becomes a measure of manager generosity rather than of performance.
Many companies between twenty and a hundred people are better served by three labels with clear consequences: on track, needs a plan, exceeds. Anything more granular tends to generate debate about the scale rather than about the work.
The performance improvement plan, done properly
A performance improvement plan is documentation of a genuine attempt to help someone succeed, and it should read that way. Specific gaps, specific support, specific measures, a defined period, and scheduled check-ins.
Plans fail when they are retrofitted to justify a decision that has already been made. Employees notice, tribunals notice, and the record becomes the company's problem rather than its protection.
A defensible plan names what will change, what the company will provide, how progress will be measured, and what happens if the measures are not met. It is signed, dated, and reviewed on the dates it promises.
Manager behaviour is the actual system
No process survives managers who avoid difficult conversations. If the top ten managers in your company cannot say the hard thing in a one-on-one, quarterly notes will be vague, ratings will drift upward, and the first honest conversation will be a termination meeting.
That is why calibration sessions and manager coaching belong in the same programme as the review cycle. The documents are downstream of the conversations, and the conversations are downstream of capability.
Questions about this
How often should performance reviews happen?
Quarterly written notes with a conversation, compiled annually into a formal review, is the rhythm that works best between 20 and 200 employees. Anything less frequent loses the record; anything more frequent becomes administrative noise.
Do we need ratings?
Not necessarily. If ratings are not tied to pay or promotion decisions, three simple labels are enough. If they are tied to compensation, they need calibration across managers and clear behavioural definitions.
Can we put someone on a performance plan immediately?
You can, but a plan that appears the week a decision was made is weak documentation. The strongest position is a pattern of documented conversations followed by a plan with support and measures.
How long should a performance improvement plan run?
Thirty to ninety days depending on the role and the gap. It should be long enough for a genuine attempt and short enough that the person is not left in limbo for a quarter without clarity.