Most performance reviews change nothing. A manager fills in a form once a year, the employee reads it, both agree it was "fine," and the same problems show up next quarter. That is a waste of everyone's time. A review framework earns its keep only when it changes what someone does the following Monday.

For operations teams the stakes are higher than in most functions, because the work is measurable. Output per shift, error rates, on-time delivery, and cost per unit are all sitting in the data. That is an advantage and a trap. The advantage is you can ground feedback in facts instead of opinion. The trap is you start rating whatever is easiest to count and ignore what actually makes an operation good — judgement under pressure, fixing root causes, and making the people around them better.

This guide covers the parts that matter: choosing the right metrics, setting a cadence, structuring the conversation, rating fairly, and turning it all into a development plan people act on. It is written for the person who owns the operation, not for the HR binder.

What a performance review is actually for

A review does three jobs at once: it tells someone honestly where they stand, it agrees what changes next, and it creates a record you can defend if a promotion or a termination is ever questioned. If your process does the first two, the third takes care of itself.

Weak looks like a backward-facing report card. The manager recites what happened, the employee nods, and nobody leaves with a clear next action. Strong is forward-facing. Most of the conversation is about the next quarter — what the person will do differently, what support they need, and how you will both know it worked. A useful test: after the meeting, could the employee write down three specific things they will change? If not, the review failed regardless of the paperwork.

To do this well, separate the review from the raise. When money is on the table people defend their score instead of listening. Run the development conversation, then handle compensation in a distinct discussion a week or two later.

Choose metrics that reflect the job, not just what is easy to count

The fastest way to break an operation is to reward one number and forget the trade-offs. Push throughput without watching quality and you get more units and more rework. Push cost-cutting without watching safety and you get a cheaper line and an incident. Every operational metric has a shadow, and a good scorecard pairs each one with the metric it can quietly damage.

Pick three to five metrics per role. More than that and none of them drive behaviour. Balance them across output, quality, cost, and people so nobody can win by sacrificing one dimension for another.

Where an established measure exists, use it rather than inventing your own. Overall Equipment Effectiveness (OEE) already blends availability, performance, and quality into one figure for equipment-heavy operations. NPS and CSAT are the standard reads on customer sentiment. Leaning on recognised measures makes your numbers comparable across teams and over time. If you are still deciding what to track, a structured look at your operations metrics will save you from measuring motion instead of results.

Set a cadence that catches problems while they are small

An annual review is a post-mortem. By the time you deliver it, the moment to fix anything passed months ago. The point of frequency is not more paperwork — it is shortening the gap between a problem appearing and someone hearing about it.

The workable rhythm is layered. Short, cheap check-ins keep the work on track; heavier reviews step back and look at the arc.

Weak cadence saves everything for the annual meeting, so nothing is a surprise except the score, which is a surprise to everyone. Strong cadence means the annual review contains zero new information — every point in it has already been discussed at a check-in. A senior operations leader who protects a weekly fifteen minutes per direct report rarely has to deliver a shock rating, because small corrections happen continuously — a habit that fits naturally into a well-run COO or operations leader's routine rather than a once-a-year event bolted on top.

The four things every review should assess

Reviews drift into vague adjectives — "great team player," "needs to step up" — unless you give them structure. Assess four distinct things and rate each on its own:

Results against goals. Did they hit the objectives you agreed? OKRs work well here because they force a measurable target upfront, so the review is a lookup, not a debate. If you are new to structured goals, SMART criteria (specific, measurable, achievable, relevant, time-bound) are the simplest place to start. Technical competence. The role-specific skills — can they run the process, read the data, troubleshoot the equipment? A supervisor might hit their numbers this quarter yet still have a skills gap that will bite when conditions change. Behaviour and collaboration. How they get results, not just whether they got them. A person who hits targets by burning out their team is a liability, and this is exactly where a numbers-only review goes blind. Tie every behavioural rating to a specific observed example, never a general impression. Trajectory. Are they growing, plateaued, or slipping? This is what turns a review into a talent development conversation instead of a scoreboard.

For roles where teamwork and leadership matter, add structured 360-degree feedback — brief, specific input from peers, the manager, and direct reports. Keep it to a few pointed questions, because vague 360s produce vague reviews.

Rate on a scale people trust — then calibrate

A rating scale is only worth using if a "4" from one manager means the same as a "4" from another. Most five-point scales fail this because every manager anchors differently — one is generous, one is harsh, and the score tells you more about the rater than the rated.

The fix is calibration: before ratings are final, managers sit together and defend their scores against evidence. If two people rated as "exceeds expectations" clearly performed at different levels, one rating changes. Calibration is the single most effective step for fairness in the whole framework, and it is the one most often skipped because it takes a couple of hours. Those hours are cheap next to a promotion that goes to the person with the most generous manager rather than the best performer.

Define each level in concrete terms, not adjectives. "Meets expectations" should read like "consistently delivers agreed targets with normal support" — a description a third party could apply, not a feeling. Data-grounded ratings also hold up better if a decision is ever challenged, which is why anchoring reviews in real numbers matters as much as the conversation itself.

Turn the review into a development plan people act on

The review is the diagnosis. The development plan is the prescription, and it is the part that actually pays back the time you spent. A plan that says "improve communication" is worthless. A plan that says "lead the Tuesday shift handover for the next quarter; we review how it went in April" is something a person can act on and you can check.

Good plans are specific, owned, and time-boxed. Name the skill, the concrete action, the support (a mentor, a course, a stretch assignment), and the date you will look at it again. Two or three real commitments beat a list of ten aspirations nobody revisits. When development plans connect to visible advancement paths, they also do quiet work on retention, because people who can see a way up are far less likely to leave — the practical core of most employee engagement work.

Keep bias out with evidence and consistent standards

Bias in reviews is rarely deliberate. It creeps in through recency (the last month outweighs the other eleven), the halo effect (one strong trait colours everything), and similarity bias (managers rate people like themselves higher). The defences are structural, not a matter of trying harder to be fair.

Keep a running log of specific events through the year so the review draws on twelve months, not the last three weeks. Require an example for every rating, run the calibration sessions described above, and apply the same written standards to everyone in the same role. None of this is about mistrust of managers — it protects good decisions from predictable human error.

Adapt the process for remote and distributed teams

Distributed operations lose the informal signals a manager picks up walking the floor, so the framework has to carry more of the load deliberately. Lean harder on outcome metrics that do not depend on being seen, and be careful not to reward visibility (who is loudest on chat) over results. Use asynchronous written input for 360 feedback so people in other time zones contribute properly instead of being an afterthought. Hold the actual review conversation live on video — a rating delivered over text lands badly and leaves no room to talk it through. The same discipline that makes remote operations management work — clear metrics, written context, deliberate check-ins — is what makes remote reviews fair.

Key takeaways

• A review is worth running only if it changes what someone does next; make most of the conversation forward-looking, and separate it from the pay discussion.
• Pick three to five metrics per role and pair each output measure with the quality, cost, or safety metric it can quietly damage. Use recognised measures like OEE, CSAT, and NPS where they fit.
• Run a layered cadence — weekly check-ins, quarterly progress reviews, an annual summary — so the annual review contains no surprises.
• Assess results, technical skill, behaviour, and trajectory as four separate things, each tied to a specific example.
• Calibrate ratings across managers before they are final; it is the single most effective fairness step and the one most often skipped.
• End every review with two or three specific, time-boxed development commitments, not a wish list.

Frequently asked questions

How often should operations teams have performance reviews? Run a layered cadence rather than a single annual event: short weekly or biweekly check-ins for course-correction, a quarterly progress review on metric trends and goals, and one annual review that summarises the year and sets a rating. The frequent touchpoints are what keep the annual review free of surprises — every point in it should already have been discussed. What is the most common mistake in operations performance reviews? Rating people on whatever is easiest to count and ignoring the trade-offs. If you reward throughput alone, quality and safety slip; if you reward cost-cutting alone, you get a cheaper operation and more incidents. Pair every output metric with the quality or safety measure it can quietly damage, and cap the scorecard at three to five metrics so each one actually drives behaviour. How do you keep ratings fair across different managers? Calibration. Before scores are final, managers meet and defend their ratings against evidence, and out-of-line scores get adjusted so a "4" from one team means the same as a "4" from another. Support it by defining each rating level in concrete, observable terms rather than adjectives, and by requiring a specific example behind every score. Should performance reviews be tied to pay? Reviews should inform pay, but do not hold both conversations at the same time. When money is on the table people defend their rating instead of hearing the feedback, and the development discussion gets lost. Run the review first, then handle compensation in a separate conversation a week or two later once the feedback has landed. What role does the operations leader or COO play in the review process? They own the framework rather than every individual review — setting which metrics matter, ensuring the process aligns with company strategy, and running or overseeing calibration so standards stay consistent across teams. They also model the behaviour by grounding their own feedback in data, which is far easier when the operation already tracks the right success metrics. What documentation should you keep for performance reviews? Keep a running log of specific events through the year, the metric data behind each rating, notes from every check-in, and the agreed development plan. This gives the review twelve months of evidence rather than the last three weeks, and it becomes your defence if a promotion or termination decision is ever questioned.