Short answer
Evidence-based feedback connects an observable action to a verifiable result. It helps the reader understand what happened, why it mattered, and what to repeat or change without turning a review into a personality judgment.
The test for every sentence
Ask whether another reviewer could inspect the evidence and reach the same conclusion. If the sentence could apply to anyone, add the action, context, and result.
Use this structure: In [situation], the employee [observable action], which led to [verified result]. [Artifact or feedback] supports the observation.
- Situation and relevant expectation
- Observable action
- Verified effect
- Source of evidence
Rewrite trait language
Trait labels and judgments about style can hide bias. Replace them with behavior that affected the work.
Hours and enthusiasm do not prove impact. Assess the result, decision quality, handoff, or other outcome.
Before and after
Their communication needs improvement.
The notes from [review] did not identify the decision or owner, so [group] needed follow-up before acting. For the next review, they will record both before the meeting ends.
Cover the period
Do not let the last two weeks stand in for the quarter. Pull evidence from across the cycle.
A recent miss does not erase months of solid work, and a recent win does not erase a pattern.
Apply the same standard to all feedback
Positive feedback needs evidence too. Replace general approval with the action worth repeating and the effect it had.
Development feedback adds a concrete next step. Example: [artifact] omitted [needed detail], which caused [observable effect]. Next time, add [specific change] before [checkpoint].
Use the writer as a cleaner, not a novelist
On rewrite, every concrete fact must stay. The tool should not add examples you did not provide.
Thin context should yield an honest, generic-but-usable draft, not a fake case study.