Running a performance cycle

What this achieves

A performance cycle is one round of reviews across a group of people, run to a calendar. This guide sets out its stages in order and marks the points where a decision becomes permanent.

The shape of it

A cycle is assembled, launched, worked through phase by phase, and finalised. Each of those verbs matters: assembling is free to redo, launching fixes the cohort, and finalising writes permanent records and shows employees their results.

Nothing in performance reopens. A submitted review is locked, a locked calibration is the agreed outcome, and a finalised cycle cannot be reversed. Plan the cycle knowing that the product will not let you take a step back.

1. The rating scale exists first

A cycle names the scale its reviews are rated against, so create a rating scale before you create the cycle. This is the decision with the longest reach: the scale is what every rating in the round means, and ratings from different scales do not compare.

Setup Considerations: rating scales is the page to read before choosing how many points your scale has and what they are called.

2. Goals are in place

Where the cycle rates goal outcomes as well as overall performance, the goals have to exist before the review opens. Create a goal or OKR covers that, and a cycle can be configured either to show reviewers the goals or to keep the two apart.

Setting goals during a review round is too late for that round. If goals are thin this time, run the cycle on overall performance and fix the goals for the next one.

3. The cycle is built

Create a review cycle assembles everything: the review period, the scale, the self and manager review forms, the participant scope, and the phase dates.

Decide the shape here rather than after launch — Setup Considerations: review cycles covers which phases are worth running and how long to give each one. Creating a cycle changes nothing for real people; launching it does.

4. The cycle launches, and the cohort is fixed

Run a review cycle starts with launching, which creates a participant record for everybody in scope and activates the first phase.

The scope is resolved at that moment and not again. Somebody who joins the company after launch is not swept into the running cycle. If a joiner has to be included, they have to be in scope before you launch, which in practice means launching after the intake rather than before it.

5. Employees write their self reviews

Employees complete their self review in the first phase. A submitted self review cannot be edited, and the screen says so before submission.

Tell people that in advance. The most common complaint about self-assessment tooling is discovering the lock after the fact.

6. Managers review their teams

Managers then review their team. Manager reviews lock on submission too, and they also lock when the cycle advances past the manager review phase — a late submission is not possible once the phase has moved on.

Advancing a phase with submissions outstanding is a real decision. Humavera offers to skip the people who have not completed, and taking that option means those employees reach the end of the cycle without that phase’s review. It is often the right call for a large cohort held up by a few, but it is a choice about named people rather than a way to clear a warning.

7. Calibration aligns the ratings

Where you calibrate, run a calibration session before finalising. Calibration adjusts manager ratings so a rating means the same thing across a cohort, and every adjustment carries a reason.

Two things are worth knowing before you start. Locking the session is what makes the outcome final, and an employee whose rating was changed at calibration sees that it was adjusted — the change is not invisible to them. A manager who has already told their report a number should know that before the session, not after.

8. Finalising releases the results

Completing the cycle writes permanent rating records and releases results to employees. Until that moment, manager ratings and comments are not visible to the people they are about; after it, they are, including any calibration adjustment.

It cannot be undone, so complete and lock calibration first. Cancelling is not the safety net it sounds like: a cancelled cycle writes no ratings at all and produces no history, which suits an abandoned round rather than a delayed one.

9. What follows the release

Results feed what happens next rather than sitting in the cycle. Ratings inform a compensation cycle where you run merit against performance, a weak outcome may lead to a performance improvement plan, and a strong one often starts a development plan or a conversation about career aspirations.

What happens next

Between formal rounds, performance carries on in feedback and 1:1s rather than in cycles — the check-in cadence is what makes the next cycle a summary rather than a surprise. Where you want a wider view than a manager’s, a 360 feedback cycle runs on its own schedule alongside this one.