The core rule for review prompts: better writing starts with better evidence.
Start with an evidence pack
Before using a prompt for review work, collect an evidence pack. This gives the draft something solid to stand on.
The evidence pack should include:
- the employee’s goals or expectations for the period
- concrete examples of work delivered
- observed strengths
- recurring friction points
- the level or role expectations that matter
- any relevant themes from prior feedback
If a manager cannot assemble that pack, the prompt should not move forward. The answer will sound finished long before it is trustworthy.
Workflow 1: evidence to draft feedback
The first useful prompt in this workflow turns real observations into a draft review section.
Prompt template:
Turn these performance observations into draft feedback.
Role level: [level].
Review period expectations: [expectations].
Evidence pack: [paste notes].
Return:
1. strengths supported by evidence,
2. growth areas supported by evidence,
3. impact on team or work,
4. one short next-step suggestion for each growth area.
Keep the tone direct and professional.
Do not invent examples. If evidence is weak, say what is missing.How to use it:
- Input: the evidence pack and level expectations
- Next step: remove anything that sounds stronger than the evidence
- Expected outcome: a draft that organizes the review more cleanly than the manager’s raw notes
The review gets better because the structure is clearer, not because the prose is prettier.
Workflow 2: draft feedback to calibration prep
Many review cycles go off course during calibration. Different managers describe similar performance in different ways. One person sounds harsh. Another sounds generous. A calibration prompt can help surface gaps before the meeting.
Prompt template:
Review this draft performance feedback for calibration prep.
Role level expectations: [expectations].
Draft feedback: [paste draft].
Return:
1. claims that are well supported,
2. claims that need stronger evidence,
3. places where the tone sounds stronger or softer than the evidence suggests,
4. questions a calibration group should resolve.
Do not rewrite the review yet. Diagnose it first.How to use it:
- Input: the draft review text and the role expectations
- Next step: use the diagnosis to improve the manager’s draft before the calibration meeting
- Expected outcome: fewer surprises when the team compares review language
This is a strong use of prompting because it asks for diagnosis rather than decoration.
Workflow 3: review draft to a development plan
A good review should not end at explanation. It should point toward next-quarter behavior. AI can help turn the review into a modest development plan that still feels tied to the evidence.
Prompt template:
Create a short development plan from this review draft.
Role level: [level].
Main strengths: [list].
Main growth areas: [list].
Return:
1. one development priority at a time,
2. the behavior to strengthen,
3. a practical way to practice it,
4. one manager support action,
5. one checkpoint for the next review cycle.
Keep the plan realistic for one quarter.How to use it:
- Input: the approved review themes
- Next step: cut anything that feels like a vague aspiration
- Expected outcome: a follow-through plan that makes the feedback easier to use
This is where a review becomes more than a documentation exercise.
The fairness checks that matter
Managers and HR partners should run a short fairness check before anything is shared.
Ask:
- Is every major claim tied to real evidence?
- Would another reviewer understand what the employee actually did well or poorly?
- Did the draft separate impact from personality judgments?
- Did the model amplify one weak example into a larger story?
- Does the development section name actions instead of vague virtues?
These checks matter because performance language carries consequences. A graceful sentence is useless if it distorts the record.
What review prompts should never hide
There are three warning signs that a review prompt is doing the wrong job:
- it turns vague notes into confident claims
- it smooths over a hard conversation the manager still needs to own
- it makes a weak review sound complete enough to send
If any of those happen, the issue is usually the input, not the model’s wording.
Save the evidence standard with the prompt
A reusable review prompt should always travel with an evidence standard. Store:
- what counts as evidence
- what does not count as evidence
- what level expectations apply
- who reviews the draft before it is shared
That way the prompt remains attached to the real review system instead of floating as a writing shortcut.
The useful test for any review prompt is simple: if the evidence pack disappeared, would the review still hold up? If not, the draft is not ready. Use prompts to organize observations into clearer feedback, sharper calibration questions, and practical next-quarter actions, but keep the manager responsible for the judgment and the conversation. In performance reviews, better structure matters only when it protects accuracy, fairness, and follow-through.