AI Prompts for Performance Reviews

A manager at a growing tech company wrote twelve performance reviews in one sitting the night before they were due. He submitted them all. The next morning, HR sent him a message: three of the reviews had an identical paragraph about "communication skills" that had clearly been copy-pasted. It wasn't dishonest — he just ran out of specific things to say and defaulted to generic filler under pressure.

The problem wasn't his intent. It was that he had no structured process for pulling specific observations together before sitting down to write. These ai prompts for performance reviews build that process.

What is a Good Performance Review?

A useful performance review does three things: it accurately reflects the employee's actual contribution over the review period, it gives specific feedback they can act on, and it creates a shared understanding of what "better" looks like going forward. Most reviews accomplish one of these three. The best ones do all three in under two pages.

AI helps not by writing the review for you — the specific observations have to come from you — but by structuring those observations into language that's precise, fair, and actionable rather than generic and forgettable.

Why It Matters

Performance reviews have unusually high stakes for short documents. A review that's too vague to be useful ("great team player, should work on communication") gives the employee nothing to act on. A review that's too critical without evidence damages trust. A review with inconsistent standards across the team creates legal and morale risk.

Getting reviews right matters for the employee receiving them and for the manager's credibility as someone whose feedback is worth listening to.

The Observation Capture Prompt

Run this before writing any review:

I'm writing a performance review for [role] who reports to me. The review period is [timeframe].

Help me capture observations by asking me 8 structured questions — one at a time — that will surface: specific examples of strong performance, specific examples of where expectations weren't met, growth over the review period, and honest assessment of potential going forward. 

Ask each question and wait for my answer before proceeding to the next.

What this does: Extracts specific observations from your memory before the writing starts. The "one at a time" instruction is critical — sequential questions prevent you from giving vague answers and moving on. It's a structured interview with your own memory.

⚡ Pro tip: Run the Observation Capture prompt at least 48 hours before you write the review, not the day of. The distance between capture and writing helps you distinguish between observations that feel significant in the moment and the ones that are genuinely representative of the whole period.

The Review Draft Prompt

Once you have your observations:

Here are my observations about [employee name/role] during [review period]:

[paste your captured observations]

Write a performance review that:
1. Opens with a clear overall assessment (1 paragraph, honest and specific)
2. Covers 2–3 areas of strong performance with specific examples
3. Addresses 1–2 areas for development with specific, actionable guidance
4. Closes with a forward-looking statement about the next review period

Avoid generic phrases like "team player," "communication skills," and "works hard." Replace every generality with a specific observation. Tone: professional, honest, constructive.

What this does: Converts your raw observations into a structured draft. The "replace every generality" instruction is the key to avoiding the copy-paste failure mode from the story above — it forces specificity at the drafting stage.

⚠️ Common mistake: Letting the AI draft a review with no specific observations and then editing it into shape. A draft built on generic language produces more generic language when edited. Start with specific observations; the draft quality depends on it.

The Self-Evaluation Prompt

For employees writing their own self-evaluations:

I'm writing a self-evaluation for [role] for the period [timeframe]. Here are things I worked on and accomplished: [list your work]. Here are areas where I faced challenges: [describe].

Write a self-evaluation that: opens with my most significant contribution and why it mattered, addresses my development areas honestly without underselling myself, describes what I've learned, and articulates one specific goal for the next period. Avoid false modesty and avoid overclaiming — calibrate to what I've actually done.

What this does: Solves the self-evaluation's twin failure modes: the employee who writes everything they did without a point of view, and the employee who writes a modest list because they're uncomfortable advocating for themselves. The "avoid false modesty and overclaiming" instruction finds the calibrated middle.

⚡ Pro tip: Before writing your self-evaluation, ask your manager: "What's the one thing you most want to see in my self-eval?" Their answer tells you what they'll actually read and what they're using the document for — which is often different from what the HR template asks for.

The Calibration Check Prompt

After drafting reviews for multiple employees:

Here are reviews I've drafted for [number] employees on my team: [paste or summarize each]. Check for: (1) whether I'm using consistent standards across all reviews, (2) any employee where my review is unusually vague compared to the others, (3) any pattern in my feedback that might reflect a bias rather than a performance difference (e.g., consistently shorter reviews for certain team members), (4) any review where my overall assessment doesn't match the specific examples I've given.

What this does: Applies a consistency check across your full set of reviews — something most managers never do. The "assessment doesn't match the examples" flag is particularly useful: if you've written "exceeded expectations" but all your examples are average contributions, something is off in your calibration.

Three Real-World Scenarios

HR manager at a 150-person company: Used the Observation Capture prompt to build a structured review process for managers who previously submitted inconsistent, often vague reviews. Required all managers to complete the prompt before submitting drafts. Average review quality, as rated by employees in surveys, improved noticeably in the following cycle.

Senior individual contributor writing a self-evaluation: Used the self-evaluation prompt after struggling for two hours with a blank page. The draft took 15 minutes and required about 20 minutes of editing — compared to the 4 hours she'd estimated for the manual version.

Team lead at a startup writing their first batch of reviews: Used the Calibration Check prompt and discovered they'd been significantly more specific and positive in reviews for employees they socialized with outside work — a bias they hadn't consciously noticed.

Common Mistakes

Writing from memory without structured observation capture. Recent events dominate memory. An employee who had a rough October but an excellent November often gets reviewed on November. The Observation Capture prompt forces a systematic look at the full period.

Generic positive language as a substitute for specific praise. "John is a great collaborator" communicates nothing. "John restructured the Q3 client reporting process in a way that saved the team approximately 3 hours per week" is specific, evidenced, and meaningful.

Skipping the development section to avoid uncomfortable conversations. A review with only praise is a missed opportunity for the employee and a credibility loss for the manager. Honest, specific development feedback — delivered constructively — is the most useful thing a review can contain.

Conclusion

Performance reviews are not primarily documentation exercises. They're one of the highest-use development conversations a manager has with each employee per year. Getting them right is worth the time it takes to structure them properly.

The prompts above don't write the review for you — your observations are irreplaceable. They structure those observations into something specific, fair, and useful. Save them in PromptABCD and run the Observation Capture well before each review cycle so you're never writing from a blank page the night before the deadline.

The Peer Feedback Integration Prompt

Many review systems include peer input. Integrating it well is harder than collecting it:

Here is peer feedback I've collected for [employee]: [paste feedback verbatim]. My own observations are: [describe]. Identify: (1) where peer and manager observations align — these are the most reliable signals, (2) where they diverge significantly — this is where I need to be most careful in my interpretation, (3) any pattern in the peer feedback that I should verify before including in the review.

What this does: Prevents peer feedback from either dominating the review or being ignored — the two most common failure modes of 360 feedback systems.

⚡ Pro tip: When peer feedback contradicts your direct observation significantly, don't average them. Investigate instead. "Five peers said X; I observed Y" is a more useful review statement than "generally, there are mixed views on X." Specificity produces clarity.

Storing Your Review Process

Performance review season happens once or twice a year — long enough between cycles that the prompt templates you used last time are hard to find. Store your full review toolkit in PromptABCD: the Observation Capture prompt, the Review Draft prompt, the Self-Evaluation prompt, and the Calibration Check. Having the system ready before cycle starts means you spend your review energy on the observations, not on reconstructing the process.

⚡ Pro tip: After each review cycle, run: "What did my reviews this cycle consistently do well and consistently miss?" That one retrospective, applied to your review writing over time, compounds into substantially better feedback each year.