Key takeaways
- Reviews should rarely surprise anyone. If they do, feedback is missing during the year.
- Clear criteria for each role matter more than the format or the software.
- Separating the development conversation from the pay conversation makes both work better.
- Calibration catches inconsistent ratings between managers before they affect pay.
- Documentation should be specific, job-related, and consistent, because reviews often become evidence.
A performance review is a structured assessment of how well an employee met the expectations of their role over a set period, usually combining self-assessment, manager assessment, and sometimes peer input. It informs development, recognition, pay, and promotion decisions.
It is the second week of December. A manager with nine direct reports has twelve days to write nine reviews, and she is staring at a blank form that asks her to rate "communication" on a five-point scale. She remembers the last six weeks clearly. January through October is a blur. She gives most people a 4, writes two sentences each, and moves on.
Her best engineer gets the same 4 as someone who barely kept up. He notices. In February he takes a recruiter's call.
Most review processes fail in exactly this way: a once-a-year form, vague criteria, and managers who are asked to judge without the notes or training to do it well. This guide covers how to design reviews that people trust.
Decide what your reviews are for
Reviews try to do several jobs at once: give feedback, plan development, document performance, decide pay, and identify promotion candidates. Those goals pull in different directions. An employee who knows a review drives their raise will understandably focus on defending their rating rather than discussing what to improve.
Before you change the form, decide which jobs matter most for your company right now. A 40-person startup might care most about development and clarity. A 2,000-person company with a merit budget needs consistent ratings it can defend. Different purposes lead to different designs.
Choosing a review cadence
Annual reviews alone are rarely enough. Memory fades, priorities change, and problems sit unaddressed for months. That doesn't mean annual reviews have to disappear.
| Cadence | Works well for | Watch out for |
|---|---|---|
| Annual review only | Stable roles, small teams with frequent informal feedback | Recency bias and surprises |
| Semiannual reviews | Most midsize companies | Twice the admin unless the form is light |
| Quarterly check-ins plus an annual review | Fast-changing goals, growing companies | Check-ins becoming box ticking |
| Continuous feedback with no formal rating | Teams with strong managers and simple pay structures | Weak documentation when pay or termination decisions come up |
A good default for growing companies is a lightweight quarterly check-in with three questions, and a fuller review once or twice a year that informs pay.
Setting criteria people can actually use
"Communication: 1 to 5" invites guesswork. Good criteria describe observable behavior and results for each level of each role.
Start with two or three outcomes that define success in the role, such as quota for a salesperson or ticket resolution quality for a support agent. Add a small set of behaviors that matter at your company, written as examples. For instance, "shares project risks early, before deadlines are at risk" is something a manager can observe. "Good communicator" is not.
If you have a career framework with levels, use it. Assessing someone against the expectations for their level is far more consistent than comparing them to coworkers.
Rating scales: three points, five points, or none
Five-point scales are familiar, but most managers cluster ratings in the middle, which makes the extra points meaningless. Three-point scales such as "below expectations, meets expectations, exceeds expectations" are easier to apply consistently. Some companies drop ratings entirely and write narrative assessments, which works when pay decisions are simple, but it makes calibration and pay equity analysis harder.
Whatever you choose, define each point with examples, and tell managers what share of a typical team tends to land at each level, without turning it into a forced curve.
Writing feedback that helps
Good feedback is specific, based on observed behavior, and connected to impact. Compare these two lines about the same project:
- "Great job on the benefits rollout."
- "You caught the carrier file error two days before open enrollment closed and fixed it with the broker, which kept 140 employees from being billed for the wrong plan."
The second tells the employee exactly what to keep doing. The same principle applies to critical feedback. "Needs to be more proactive" is vague. "On the Q2 vendor project, the delay surfaced at the deadline instead of two weeks earlier when you first saw it" is something the employee can act on.
Encourage managers to keep a simple running note for each direct report during the year. Ten minutes a month removes most of the December scramble.
Common rating biases and how to reduce them
Every manager carries some bias into reviews. The common ones have names:
- Recency bias: the last few weeks outweigh the whole period.
- Halo and horns effects: one strong or weak trait colors every rating.
- Leniency and central tendency: everyone gets good or average ratings to avoid hard conversations.
- Similarity bias: managers rate people who remind them of themselves more favorably.
- Proximity bias: people who are seen in the office get credit that remote colleagues doing the same work don't.
Specific criteria, running notes, structured self-reviews, and calibration meetings all reduce these effects. So does asking managers to cite one example for every rating they give.
Running calibration meetings
Calibration is a meeting where managers review proposed ratings together, compare them against shared standards, and adjust ratings that are out of line. Done well, it makes ratings fairer and helps newer managers learn what "exceeds" looks like.
Keep calibration focused on evidence. Each manager briefly explains ratings at the edges, such as top and bottom ratings, with examples. HR or a senior leader facilitates and watches for patterns, such as one team rating everyone high, or ratings that differ by gender, race, or work location. Document changes and the reasons for them.
Holding the review conversation
The written review matters less than the conversation. A structure that works:
- Ask the employee how they think the period went before sharing your view.
- Walk through strengths with specific examples.
- Cover one or two areas to improve, also with examples. More than two rarely sticks.
- Agree on goals and development for the next period.
- Ask what the employee needs from you.
Hold pay discussions separately when you can, even a week later. When salary is on the table, people stop listening to feedback.
Connecting reviews to pay without breaking either
Most companies link merit increases to ratings, often through a merit matrix that considers both performance and where someone sits in their salary range. That keeps raises consistent and defensible. Just make sure ratings are calibrated before they feed pay decisions, and review outcomes for unexplained differences across groups. Our guide to salary bands and compensation strategy explains how merit matrices work.
Self-reviews worth reading
Most self-reviews are either modest to the point of being useless or written like a sales pitch. A few good prompts fix that.
Ask employees to list their three most important results for the period, with evidence. Ask what got in their way. Ask what they want to learn next. And ask one hard question: "What would you do differently if you ran this period again?" People answer that one honestly more often than you would expect, and it gives the manager an easy opening for a real conversation.
Give people the self-review a week before the manager review is due, not the same day. A rushed self-review is just a list of projects.
Peer and 360 feedback: when it helps
Peer feedback is valuable for roles that work across teams, where the manager sees only part of the work. It also catches problems a manager can miss, such as someone who is pleasant upward and difficult sideways.
It can also go wrong. Keep it short, with two or three questions, and limit each employee to a handful of reviewers chosen with the manager. Use peer comments to inform the manager's view, not to calculate a score. And be honest about anonymity: in a five-person team, comments are rarely truly anonymous, and people write accordingly.
Reviewing remote and hybrid employees fairly
Remote employees face a real risk of being judged on visibility instead of results. Base reviews on outcomes and documented contributions, ask managers to check whether office-based employees are consistently rated higher than remote ones doing similar work, and make sure remote employees have the same access to stretch assignments that lead to promotions.
Handling underperformance and documentation
Reviews often become evidence when employment ends. Specific, consistent, job-related documentation protects both the company and the employee. When performance falls short, address it as soon as it happens, not only at review time. A performance improvement plan should state the gap, the expected standard, the support you will provide, a realistic timeline, and how progress will be checked. Treat employees in similar situations consistently, and involve HR before any final decision.
Once you have more than a few managers, performance management software keeps goals, check-ins, notes, and reviews in one place. See our performance management buyer guide and the Lattice vs Culture Amp comparison.
Performance review cycle checklist
Work through this list with your team. The printable PDF version has space to tick each item off.
- Decide the main purpose of reviews this cycle
- Confirm the review cadence and calendar
- Write two or three outcome measures for each role
- Define each rating level with examples
- Share the timeline and expectations with managers
- Train managers on bias and writing specific feedback
- Collect self-reviews before manager reviews
- Require an example for every rating
- Hold calibration meetings before ratings are final
- Review rating patterns across groups and locations
- Hold review conversations before pay conversations
- Set goals and development plans for the next period
- Document underperformance specifically and consistently
Where to go from here
Wherever you are in the process, there is a next step that fits.
Compare software categories
Email templates for this topic
Ready-to-edit emails with subject lines, compliance notes, and a free PDF.
- Performance review reminder email templateLaunch the review cycle and remind managers of deadlines, expectations, and fairness checks.
- Self-review request email templateAsk employees to complete a self-assessment with clear questions and a useful purpose.
- Performance improvement plan (PIP) email templatePut a performance improvement plan in writing with clear goals, support, check-ins, and consequences.
- Training enrollment email templateEnroll employees in required or optional training with dates, time, pay, and completion deadlines.
Frequently asked questions
How often should performance reviews happen?
Should performance reviews include ratings?
What is calibration in performance reviews?
Should pay be discussed in the performance review?
What should a performance improvement plan include?
Sources
We checked the facts in this guide against these primary sources. Laws and thresholds change, so confirm current rules before you act.
Written by the HRTally Research Team
Our guides are written and reviewed by people with hands-on US HR, payroll, and HR technology experience. Legal and tax facts are checked against primary sources, such as the Department of Labor, IRS, EEOC, and USCIS, and linked at the end of each guide. Vendors never review or pay for our editorial content.