How to Write Better Performance Reviews

---

When engineering managers write all-positive performance reviews, they usually think they are protecting morale. In reality, they are destroying trust. To evaluate your team effectively, you must realize that useful candor requires evidence, not just enthusiasm.

This guide covers why generic praise reads like indifference, how to anchor reviews in shared career ladders, and why you must stop relying on your memory to assess performance.

1. The Danger of Generic Praise

The all-positive review looks kind, but it often lands as indifference.

When a review is filled with phrases like "great work" and "keep it up" but contains no specific examples, the engineer cannot tell what actually mattered. Praise without evidence stops functioning as recognition and starts reading like conflict avoidance.

If you write a review with "zero specifics," you are telling the engineer that you did not pay attention to their work. Useful candor requires evidence, not just enthusiasm.

2. Recognize the Structural Trap

Managers do not write hollow reviews because they are lazy; they write them because the system is noisy.

Research shows that up to 70% of employees are often clustered in the top two rating categories of subjective review systems. When standards are fuzzy, managers retreat into vague positivity because it feels safer than making a precise judgment they cannot defend.

The trap is structural. You cannot fix it with better prose; you must fix it with better standards.

3. Anchor to the Ladder, Not Your Memory

You cannot write a fair review if the team has never agreed on what "good" looks like.

Before you write a single word, anchor the review in your engineering progression framework.

What level are we evaluating against?

Does this engineer understand what "senior-level scope" actually means on this team?

Once the standard is clear, you must prove your judgment with evidence collected over time. If you write the review from memory, you will fall victim to recency bias. You must collect artifacts - design docs, PRs, and incident reports - to justify your rating.

4. Separate Recognition from Evaluation

One reason all-positive reviews fail is that they blur three different jobs into one paragraph.

The Strong Structure:

• Recognition: Name the real contributions (e.g., "You improved the deployment pipeline reliability").
• Evaluation: Compare the work against the ladder (e.g., "You are operating consistently at the L3 level").
• Development: Define the next step (e.g., "The next step is broader technical influence").

Do not collapse these together. When praise does too much work, it ends up doing none of it well.

5. Calibrate to Remove Bias

Engineering managers often hate calibration, viewing it as HR theater. But without it, your review is just an isolated opinion.

Calibration is an operational control. It forces you to defend your "all-positive" review to your peers. If your review is grounded in role expectations and concrete evidence, it will survive calibration. If it is grounded in vague enthusiasm, your peers will spot the inflation immediately.

6. Avoid the "Brutally Honest" Overcorrection

The opposite of an all-positive review is not a harsh review. It is a precise one.

A meta-analysis of over 600 feedback interventions found that more than one-third actually decreased performance, usually because the feedback was ego-threatening or vague.

Do not mistake bluntness for clarity.

Your goal is not to intensify the emotional charge of the conversation; your goal is to reduce ambiguity. Useful candor requires evidence, not just enthusiasm.

Closing Thoughts

The all-positive review is a symptom of a manager who is unsure of their own judgment.

If you want to build trust, stop using positivity to cover your uncertainty. Start using evidence to clarify your expectations.

If an engineer cannot reconstruct your judgment from the examples you provided, your review needs more than better phrasing. It needs a better operating model.

Do This Next: The Review Calibration Checklist

Audit your draft performance reviews against these four items.

• [ ] The Evidence Test: Does every paragraph of praise include a reference to a specific project, PR, or incident?
• [ ] The Ladder Link: Have I explicitly stated how the engineer's work maps to the expectations of their current level?
• [ ] The Edge Case: Have I identified at least one specific "growth edge" or area for improvement for the next cycle?
• [ ] The Memory Check: Did I pull examples from the entire review period, or just the last month? (If just the last month, go check Jira).