Publish the calendar and the criteria before the cycle opens, confirm goals are current, collect self-assessments, then manager assessments, then calibrate across managers. Decide pay and promotion separately from the development conversation. Hold the delivery conversations, document outcomes in the employee record, and close with a retrospective on the cycle itself.
Publish the calendar with every date on it: when self-assessments close, when manager assessments close, when calibration happens, and by when conversations must be held. Settle eligibility next, covering who is in scope, how people who joined part-way through are treated, what happens to someone on long leave, and who reviews a person with a dotted-line manager. Settle the rating scale and write out what each point means in words, because a scale without anchors becomes whatever each manager privately thinks it means. Then communicate all of it once, in writing, to managers and employees together. Ambiguity discovered in the third week of a cycle costs far more than the week spent settling it in advance.
Because assessing someone against goals set a year ago and overtaken by a reorganisation is an argument waiting to happen. Ask every manager to review their team's goals for currency: which still stand, which were changed by circumstance, which were withdrawn, and record the change with a date rather than editing history silently. Where goals were never set properly, say so in the record and assess against role expectations instead, rather than inventing goals retrospectively. This step is unpopular and it decides whether the cycle is defensible later, when somebody disputes a rating or a promotion outcome and asks what exactly they were measured against.
Managers meeting to compare how they have applied the same scale, before ratings are final and before anyone has been told anything. The purpose is consistency, not a forced distribution: one manager who rates generously and another who rates severely create unfair pay outcomes for people doing comparable work. Run it in groups small enough for real discussion, with a facilitator who is not among the managers being calibrated, and require evidence for outliers at both ends. Record what changed and why. Where a rating moves, the manager who owns the relationship delivers the outcome and owns the explanation, which means they need to leave the room agreeing with it.
Because nobody absorbs coaching while waiting to hear about money. Holding both in one conversation reduces it to the number, and the developmental part is heard as justification rather than help. Run assessment and calibration, hold the performance conversation, then communicate pay and promotion outcomes separately once decisions are approved. The separation also protects the integrity of the ratings, since a manager who knows a rating maps straight to a payout will rate towards the payout they want to give. Where the gap between the two conversations is long, tell people when to expect the second one, because silence after a review is read as bad news.
Booked in advance with enough time, held in private, and prepared for by both sides. The manager opens with the outcome rather than burying it after ten minutes of context, gives the evidence, then spends the bulk of the time on what happens next: what the person is developing, what support they get, what changes in their work. Disagreement should be recorded rather than argued into submission, with a stated route to escalate. For anyone rated below expectation, the conversation ends with a written plan, specific enough to be assessed, and a date to review it. A conversation with no agreed next step has not finished, it has merely stopped.
The assessment, the rating, the evidence behind it, any change made at calibration, the employee's own comments and the date of the conversation. This is the record you rely on when a decision is challenged months later, so it belongs in the employee record rather than in a manager's inbox or a shared drive somebody will tidy. Purpose-built [performance management software](/performance-management-software) keeps that trail intact and stops the cycle depending on file naming discipline, though the habit that matters is writing evidence at the time rather than reconstructing it. Set an expectation about tone as well: notes written for a private audience read very differently when they are produced in a dispute.
Close the cycle formally and ask what to change while everyone still remembers. Look at completion: which stages ran late, where the chasing effort went, which managers needed help. Ask employees whether the conversation was useful and whether the outcome surprised them, because surprise means feedback was not happening in between. Ask managers which part of the process they would drop. Then change two or three things and write them into next cycle's plan rather than into a document nobody reopens. Pulling completion and timing data from the [employee record](/hrms) keeps the retrospective factual instead of an argument about who was slowest.
Pitch N Hire is an applicant tracking system built for recruiters and hiring teams. If this answer described something you want to run properly, the ATS is where it lives.
Free for 1 user · No credit card · Talk to a real hiring expert
Get a personalized walkthrough of Pitch N Hire on your own roles and workflow. No slides, no obligation.
Prefer to talk? Book a demo · Talk to sales · View pricing
Free 1-user plan · No credit card · Talk to a real hiring expert
See your true cost-per-hire and how much Pitch N Hire could save you — our free Recruitment ROI Calculator gives you the numbers in under a minute. No signup required.
Open the free ROI calculatorPrefer a tailored walkthrough on your real roles? Drop your work email:
★ Free 1-user plan · No spam · Talk to a real hiring expert
