How to conduct a performance review that drives growth

Knowing how to conduct a performance review that improves performance starts with five actions: use specific evidence, invite the employee’s perspective, agree on one or two observable changes, document the manager’s support, and schedule a follow-up before the meeting ends.

The meeting alone changes nothing. A manager can deliver an accurate, thoughtful review and still see no improvement because the conversation never reaches the next attempt at the work.

Most managers already know they should prepare, listen, and avoid vague criticism. Effective performance reviews turn a year of evidence into a small commitment that both people can act on. This guide gives you that performance review process, including the questions, a diagnosis table, a copyable improvement plan, and a complete example.

Why most performance reviews do not change performance

An employee performance review often tries to do too many jobs at once. It records a rating, explains pay, recognizes strengths, corrects problems, discusses career goals, and meets an HR deadline. Development becomes one item in an overloaded agenda.

The emotional stakes make this harder. An employee who is trying to understand a compensation decision may have little attention left for coaching. A manager who has delayed difficult feedback may soften the message until nobody knows what should change.

Feedback itself is not reliably helpful. A landmark meta-analysis by Kluger and DeNisi examined 607 effect sizes and 23,663 observations. Feedback improved performance on average, but more than one-third of the interventions reduced it. The effects weakened when attention moved away from the task and toward the person’s sense of self. See the feedback-intervention meta-analysis.

That finding does not prove that annual reviews fail. It makes a narrower point: more feedback is not automatically better. “You are not strategic enough” invites an argument about identity. “In the last two planning meetings, your proposal described the preferred option but not its risks or alternatives” gives both people something they can examine and practise.

Use a simple improvement loop:

Start before you open the review form: write one sentence for each step about the most important development area. Any blank step shows what you still need to learn.

Prepare evidence, not a verdict

Good preparation is not writing the final judgment in advance. It is assembling enough evidence to hold a fair conversation and being willing to revise your interpretation when new context appears.

Review expectations and earlier commitments

Start with the role, goals, decision rights, and commitments from previous conversations. Ask whether the standard was clear and stable throughout the period. If priorities changed three times but the written goal did not, the old goal is weak evidence.

Look at both outcomes and how the employee produced them. A strong result achieved by hiding risk or exhausting colleagues may not be repeatable. A missed result caused by a canceled project or delayed executive decision should not become a character judgment.

Gather examples across the whole period

Recency may be convenient, but it is not a fair sample.

Use lightweight notes from one-on-ones, project retrospectives, goals, customer outcomes, work products, and prior feedback. Look for examples from different points in the review period. Do not turn routine note-taking into surveillance; record decisions, commitments, outcomes, and feedback that the employee can reasonably know exists.

For each important conclusion, write:

• The situation and relevant expectation.
• The behavior, decision, or output you observed.
• The impact on the work, customer, or team.
• Context that may have affected the result.
• Whether the employee heard this feedback earlier.

If you cannot produce an example, soften the conclusion or keep investigating. “People find you difficult” is hearsay. “During the launch review, you interrupted two owners before they could explain their risk estimates, and both stopped contributing” is a claim the employee can understand, question, and address.

Check your own pattern recognition

Before finalizing the review, ask what may have distorted your sample:

• Did recent work outweigh earlier work?
• Did you observe office-based employees more often than remote employees?
• Did one success or failure color unrelated judgments?
• Are you rewarding a work style similar to your own?
• Did you apply the same standard and evidence threshold across the team?

This check cannot remove bias. It can expose a conclusion that needs more evidence or calibration.

Ask the employee to prepare

Send a short prompt several days before the meeting:

1. Which outcomes are you most proud of, and what helped produce them?
2. What did not go as planned, and what contributed to that?
3. Which skill, behavior, or responsibility would make the biggest difference next?
4. What support or decision do you need from me?

This is not a test of whether the employee guesses your feedback. Their answers give you evidence, surface missing context, and make the meeting less like a verdict being read aloud.

Imagine a manager reviewing a product designer after two launches. The manager remembers a late handoff from last month and drafts “needs to improve reliability.” Earlier project records show that the designer delivered the first launch on time and flagged the second project’s scope risk three weeks before leadership made a decision. This hypothetical review changes once the manager looks beyond the most recent outcome: the useful discussion is about escalation and decision timing, not the employee’s reliability.

Make the next review easier: create a private, factual log for wins, concerns, commitments, and follow-ups. Two minutes after a meaningful event beats reconstructing 12 months at the deadline.

How to conduct a performance review as a diagnosis

The employee should leave knowing what you observed, what matters next, and what support exists. Getting there requires a conversation, not a polished monologue.

Set the purpose and agenda

Open with the purpose in plain language. For example:

“I want us to look at the work, recognize what is going well, understand the biggest constraint, and agree on one or two priorities for the next period. I have evidence and a point of view, and I want to test both against what you saw.”

If the meeting includes a rating or compensation decision, explain the agenda. State what has already been decided, what is open for discussion, and when there will be space for development planning. Do not pretend a fixed pay decision is a collaborative coaching exercise.

Let the employee speak before you close the story

Ask for their assessment first. Listen for evidence, not agreement. The most useful performance review questions invite examples and context:

• What result best represents your contribution this period?
• Where did the work become harder than expected?
• Which feedback have you already acted on?
• What would you do differently with the same situation?
• What part of your work do I see least clearly?

Letting the employee speak first does not mean avoiding the manager’s judgment. It reduces the risk of anchoring the entire conversation on the first conclusion voiced in the room.

Describe the work and its impact

Use a four-part statement:

1. Expectation: What result or behavior did the role require?
2. Evidence: What did you directly observe?
3. Impact: What happened to the work or other people?
4. Question: What context or interpretation might you be missing?

Here is how that sounds:

“We agreed that project risks would be raised while the team still had options. In the Atlas and Beacon projects, partners learned about material risks after the forecast had already slipped. That left Operations with less time to change staffing. What was happening from your perspective, and what made earlier escalation difficult?”

Keep personality out of the statement. “Careless,” “not leadership material,” and “bad attitude” compress many possible causes into labels the employee cannot practise their way out of.

Diagnose the blocker before choosing the response

The same missed outcome can require different action. Use this table to test the cause with the employee:

Do not force a shared diagnosis in one sitting when facts are missing. Name what each person will verify and set a date to reconvene.

Turn feedback into a performance improvement plan

An ordinary development commitment is not the same as a formal performance improvement plan (PIP). A PIP may be part of a corrective process with policy and legal implications. Follow your organization’s policy and get appropriate HR or legal guidance rather than disguising corrective action as casual coaching.

For routine development, keep the plan small. Choose one or two priorities with the greatest effect on the role. A list of eight weaknesses gives the employee eight reasons to spread their attention thinly.

Define the next successful attempt

“Communicate better” is not an action. Define the situation where behavior should change and what the employee will do.

Useful commitments sound like this:

• Send a risk note within one working day when an agreed threshold is crossed.
• Bring two options and a recommendation to the weekly decision meeting.
• Ask the customer to confirm the problem in their own words before proposing a solution.
• Give the reviewer a first draft two days before the deadline for the next three reports.

The commitment should be demanding enough to matter and narrow enough to observe.

Put manager support in the same row

Development fails when the employee receives a goal but not the conditions to practise it. Write what the manager will do: clarify a threshold, create an opportunity, review an early attempt, remove a blocker, make an introduction, or decide a tradeoff.

Research supports connecting feedback to goals, though it does not prove that every workplace plan will work. One meta-analysis comparing feedback plus goal setting with goal setting alone reported an effect of d = .63 across 16 effect sizes. Contexts varied, so use this as a reason to create a testable plan, not as a promise. See the feedback and goal-setting meta-analysis.

Copyable improvement-plan template

Schedule the first check within 30 days. Use a shorter interval when the employee will attempt the behavior sooner or when delay creates significant risk.

Turn the conversation into work: copy this table into the existing review document and complete it together before either person leaves the meeting.

Worked example: replace “communicate better” with observable practice

Picture a hypothetical project lead who sends risk updates after deadlines are already threatened. The manager plans to say, “You need to be more proactive and communicate better.” That sentence identifies frustration, not a path forward.

The manager instead brings two examples. In both projects, the lead knew about a material delivery risk several days before affected partners learned about it. The late warning reduced the options available to the team.

During the conversation, the lead explains that they do not know which risks are important enough to escalate and worry that frequent warnings will create noise. The diagnosis changes the plan. This is partly a behavior issue and partly an unclear threshold.

Their improvement plan becomes:

This hypothetical plan does not guarantee better delivery. It creates a fair test. If the lead sends clear warnings and decisions still arrive late, the next diagnosis belongs with the decision process rather than the employee’s communication.

Handle disagreement without losing the next step

An employee does not need to agree with every conclusion for the conversation to remain useful. Ask what they dispute: the event, the expectation, the interpretation, the impact, or the rating.

Review the evidence and add material context to the record. If you need more information, say so. Do not demand immediate emotional acceptance, especially when the decision affects pay or career opportunities.

Separate two questions:

1. What do we believe about the past period?
2. What will the role require from this point forward?

You may remain divided on the first and still clarify the second. Document unresolved disagreement accurately rather than rewriting it as consent.

Consider a hypothetical analyst whose rating depends partly on a report delivered late. The analyst shows that the source data arrived four days after the agreed date. The manager may still identify an improvement opportunity around earlier escalation, but should correct any claim that the analyst controlled the original delay. The next commitment becomes specific without preserving a false account of the past.

For high-stakes corrective issues, pause if necessary and follow the applicable policy. Routine development guidance should not substitute for a formal process.

Follow up until the change becomes routine

The first follow-up should review the next attempts, not ask, “How do you think it is going?” Bring the evidence named in the plan.

A 15-minute agenda is enough:

1. What attempt occurred since the review?
2. What evidence shows progress or continued difficulty?
3. What did the employee learn?
4. What feedback or support did the manager provide?
5. Should the plan close, continue, or change?
6. When is the next check?

Give feedback close to the work so the employee can apply it to another attempt. If progress stalls, revisit the diagnosis before repeating the same advice more firmly. The employee may need clearer examples, more practice, a removed blocker, or a different goal.

MIT Human Resources’ performance-review guidance treats reviews as part of ongoing dialogue. Its process includes reviewing prior commitments and results, discussing obstacles, setting goals, recording notes, and planning follow-up. That sequence is practical because it prevents the annual form from becoming the final destination.

Close the plan explicitly when the new behavior is consistent. Continuing to monitor a resolved issue forever turns development notes into a permanent penalty.

Optional video: “Conducting Performance Reviews,” a current video result for the target query. Use the process and evidence in this guide as the working standard.

Performance review mistakes that block improvement

Seven mistakes repeatedly break the improvement loop:

• Saving negative feedback for the formal review. Raise meaningful issues when they happen, then use the review to examine the pattern and progress.
• Reconstructing the year from recent memory. Use goals, outcomes, commitments, and earlier feedback from across the period.
• Using personality labels. Replace “not strategic” or “not a team player” with observable work, impact, and the expected alternative.
• Packing the plan with goals. Select the one or two changes that matter most and can be practised.
• Forcing a praise-criticism-praise sandwich. Give specific recognition because it matters, then discuss the main development issue plainly.
• Assigning all change to the employee. Put the manager’s decisions, feedback, and blocker removal in the plan too.
• Ending without a date. “Keep working on it” is weaker than a calendar invitation tied to agreed evidence.

The follow-up is where the review earns its value

An accurate review without practice and follow-through is documentation, not development. The form may be complete, but the work remains unchanged.

Treat the meeting as the point where both people choose the next experiment. Name the behavior or outcome, the manager’s support, the evidence, and the first review date. Then discuss the next attempt while it is still fresh enough to learn from.

Take one upcoming performance review and put the follow-up on the calendar before you finish writing the form. That small act forces the conversation to point toward real work rather than next year’s archive.