Loading page
All-in-one HR platform · v2.0
14-day free trial · No credit cardhello@HrMagix.com
All-in-one HR platform · v2.0
14-day free trial · No credit cardhello@HrMagix.com
Calibration exists because two managers rarely mean the same thing by the same rating. One gives a 4 to anyone who did the job well; another keeps a 4 for the person who changed how the team works. Left alone, those ratings flow into increments and promotion lists, and the employee with the stricter manager pays for it.
This guide covers the calibration meeting itself: what to prepare, how to chair it, and what to do afterwards. For what calibration means and how a 9-box grid frames performance against potential, see the calibration glossary entry and the KRAs and 9-box topic page.
01
A calibration meeting works when the group in the room knows the people being discussed. Calibrate by department or by a group of related teams, not the whole organisation at once. A manager cannot argue usefully about someone they have never worked with.
Fix three things in writing before the meeting: which ratings are being calibrated (overall rating, 9-box placement, or both), whether there is any guidance on distribution, and who has the final say if the room cannot agree. If your organisation uses a guide distribution, say whether it is a target or a hard rule. Most disputes in the room trace back to this being left vague.
02
Managers should submit provisional ratings with a short evidence note before the meeting, not bring them in their heads. The note should name two or three specific outcomes against the person's KRAs or goals, not adjectives.
HR compiles these into one pre-read: every employee with provisional rating, role, level, time in role and the evidence note, plus a view of the spread of ratings by manager. The spread view is the most useful page in the pack. It shows at a glance which manager rates high, which rates low, and where the conversation needs to go.
Watch for itCirculate the pre-read at least a day ahead and ask attendees to read it. A meeting where people read the evidence for the first time on the screen turns into a reading session.
03
The facilitator, usually HR, does not rate anyone. Their job is to keep the discussion on evidence, keep time and record decisions. Open by restating the scope and the definitions of each rating level, so everyone is comparing against the same words.
Do not walk through every employee in the same depth. Spend the time where a decision changes something for the person: the top and bottom of the scale, people near a boundary, and anyone whose rating differs from what their peers in similar roles received.
04
When managers disagree, ask for the evidence again rather than for opinions. Questions that help: what did this person deliver against their goals, how does that compare with the person rated one level higher, and would the rating hold if their manager were someone else.
Watch for the familiar distortions: the most recent quarter outweighing the year, a loud advocate carrying a weak case, a quiet manager's team being marked down, and ratings tracking how visible someone is rather than what they delivered. Name the pattern when you see it, neutrally, and move back to the evidence.
Watch for itDo not let calibration become the place where a manager first hears a concern about their own report from a peer. If new information about conduct comes up, park it and handle it separately.
05
Every rating that changes in the room needs a recorded reason in one or two lines. That record protects the decision if it is questioned later and lets the manager explain it clearly. Ratings that did not change need no note.
After the meeting, the manager, not HR, tells each employee their final rating in the appraisal conversation. A manager whose provisional rating was moved down must be able to explain why in terms of the person's work, never as "calibration lowered it". Close the loop by reviewing, at the end of the cycle, whether the definitions or the evidence notes need changing for next time.
Every guide ends here rather than in a summary. If a line below is not true yet, that is the next thing to do.
A meeting where managers who rate employees in related teams compare their provisional ratings against shared definitions and evidence, so the same rating means the same thing across managers before ratings are finalised.
The managers whose teams are being calibrated, their own manager, and an HR facilitator who keeps the discussion on evidence and records decisions. People who do not know the employees being discussed add little.
That is the employer's decision. Some use a distribution as guidance to prompt discussion, others apply it as a rule. Whichever you choose, state it before the meeting so it is not argued case by case.
In terms of the employee's work and how it compared with the rating definitions, using the reason recorded in the meeting. Saying the rating was lowered by calibration, without a reason, damages trust in the process.
After managers have submitted provisional ratings and before ratings are shared with employees or used for increments, so that changes are made once and communicated once.
The checklist above is generic by necessity. Bring your own structures, states and rota and we will tell you which lines are already true and which are not.
14-day free trial · Full access to all 12 modules · Zero implementation fee · Cancel anytime