AI Skill Pod

Management

Performance review draft

Official

Official · updated Nov 4, 2026 · v1.0.0

Drafts a performance review from the manager's evidence, per expectation with examples and impact, then checks it for rating bias and vague words.

Sign up free to add this skill to your pod

You come back here after you sign in, and add it with one click.

Sign up freeAlready have an account? Sign in

When to use it

What you give it

What you get

A structured review with dated examples, growth areas, rating rationale, and a bias-check section for the manager.

The SKILL.md

---
name: performance-review-draft
description: Drafts a written performance review from evidence the manager provides - for each expectation, specific examples, their impact, growth areas, and the rationale for the rating - then runs a bias check for recency, halo or horn effect, leniency, vague praise, and personality words. Use when a manager asks to write a performance review, annual or mid-year review, self-review response, promotion write-up, or review comments for a direct report.
---

# Performance review draft

> HR-sensitive. This drafts from the manager's evidence only. Ratings, pay and promotion follow your company's process; anything not in the input is marked "[check: ...]".

A fair review is specific enough that the person could point to the week each example came from. Vague praise and personality words are where bias hides: in more than 200 reviews from 4 companies, Stanford's Shelley Correll and Caroline Simard found 57% of women got vague praise against 43% of men, and 40% of women got feedback tied to business outcomes against 60% of men. So every statement here ties a behavior to an outcome.

## When to use

- Writing an annual, mid-year, or probation review for a direct report.
- Turning a manager's scattered notes, peer feedback, and project results into a structured draft.
- Checking an existing draft for bias and vague language before it goes to calibration.
- Not for a performance improvement plan or a termination: involve HR directly.

## What you need

- The expectations or competencies the review is against (job level guide, goals, OKRs). If missing, ask once; otherwise use the goals set at the start of the period and mark "[check: expectations not provided]".
- Evidence across the whole period: project results, examples, peer feedback, the person's self-review. Ask for at least one example from each quarter.
- The rating scale and its definitions. Never invent a scale; if missing, write "[check: rating scale]".
- Don't use demographic details, health, leave, or family status as evidence.

## Steps

1. **List the expectations.** One section per expectation or competency, in the company's own words. Rate each one separately before forming an overall view, so one strong area doesn't lift the rest.
2. **Place each piece of evidence.** For each example: what happened, when, and the outcome (customer, revenue, quality, team). Use the SBI shape: situation, observable behavior, impact. Evidence without an outcome stays in the notes, not the review.
3. **Spread the evidence over the period.** If more than half the examples come from the last 2 months, ask the manager for earlier ones. That's recency bias.
4. **Write growth areas as behaviors with a next step.** "Share the analysis plan before building the dashboard, so stakeholders can redirect early", not "needs to communicate better". Our default: 1 to 3 growth areas, each with one example and one concrete next step.
5. **Write the rating rationale.** Map the evidence to the scale's definition in 2 to 4 sentences. If the evidence doesn't support the rating the manager wants, say so in a note to the manager, not in the review.
6. **Run the bias check.** Flag and rewrite: recency (examples cluster at the end), halo or horn (one trait colors every section), leniency or severity (every rating at one end), central tendency (everything "meets"), vague praise ("great team player"), and personality words ("abrasive", "aggressive", "emotional", "bossy", "shy"). Replace each with the behavior and its effect.
7. **Write the summary last.** 3 sentences: biggest contribution, main growth area, what the next period should focus on.

## Output format

```markdown
# Performance review — [name], [role], [period]
*Draft from the manager's evidence. Ratings follow [company] process. [check: items] need confirming.*

## Summary
[3 sentences: biggest contribution · main growth area · focus for next period]

## [Expectation 1, company's wording]
**Rating:** [rating on the company's scale]
- **[Month]:** [Situation and behavior] → [outcome]
- **[Month]:** [Situation and behavior] → [outcome]
**Rationale:** [2–4 sentences mapping evidence to the scale definition]

## Growth areas
1. [Behavior to change] — e.g., [example] — next step: [specific action]

## Overall rating
[Rating] — [rationale linking the sections]

## Bias check (for the manager, remove before sharing)
- Recency: [n] of [n] examples from the last 2 months — [ok / ask for earlier examples]
- Halo/horn: [ok / section X echoes section Y without its own evidence]
- Vague or personality words found: [list → rewrite]
- Ratings spread: [ok / all at one level: check]
```

## Example

**Input:** "Review for Tomas, Account Executive, H1. Expectations: hit quota, build pipeline, collaborate with CS. Closed 108% of quota. Q1: sourced the Nordwell deal himself, $42k. May: handed off Brightpath to CS with no notes, onboarding slipped 2 weeks. He's a natural, really confident, sometimes a bit aggressive with prospects. Rating scale: below / meets / exceeds."

**Output:**

```markdown
# Performance review — Tomas, Account Executive, H1
*Draft from the manager's evidence. Ratings follow company process. [check: pipeline numbers] need confirming.*

## Summary
Tomas closed 108% of his H1 quota, including a $42k deal he sourced himself. His main growth area is handoffs to Customer Success. Next half, the focus is complete handoff notes on every closed deal.

## Hit quota
**Rating:** exceeds
- **H1:** Closed 108% of quota.
- **Q1:** Sourced and closed Nordwell ($42k) without a marketing lead.
**Rationale:** Above-quota attainment across the half meets the "exceeds" definition [check: scale definition].

## Build pipeline
**Rating:** [check: no pipeline evidence provided beyond Nordwell]

## Collaborate with CS
**Rating:** below
- **May:** Handed Brightpath to CS without handoff notes → onboarding started 2 weeks late.
**Rationale:** One documented handoff missed its basic requirement, with customer impact. [check: other handoffs in H1]

## Growth areas
1. Handoffs — e.g., Brightpath in May — next step: use the CS handoff template on every closed deal, reviewed in 1:1s.

## Overall rating
[check: overall depends on pipeline evidence and the weighting of sections]

## Bias check (for the manager, remove before sharing)
- Recency: 1 of 3 examples from the last 2 months — ok
- Halo/horn: "a natural, really confident" is a general impression → removed; ratings set per section
- Vague or personality words found: "natural", "confident", "aggressive with prospects" → removed; if prospects pushed back, add the date, what he said, and the prospect's reaction
- Ratings spread: ok
```

## Checks before you send

- [ ] Every example has a date or period and an outcome.
- [ ] No personality words remain; each was replaced with a behavior or removed.
- [ ] Examples come from across the whole period, not just the last 2 months.
- [ ] Each expectation was rated on its own evidence.
- [ ] Nothing in the review (numbers, scale, policy) is invented; gaps are marked "[check: ...]".
- [ ] The bias-check section is marked for removal before sharing.

## Common mistakes

- Copying the self-review's adjectives. Use its examples, not its labels.
- Writing a rating that the evidence doesn't support, then padding the text to justify it.
- Mentioning leave, health, or personal life. They are not performance evidence.

Download SKILL.md

Changelog

Sources

  1. Another Roadblock for Women: Performance Reviews - CFO.com (2026-11-04)
  2. Beware of Rater Errors! - UTHSC Human Resources (2026-11-04)
  3. Use SBI (Situation-Behavior-Impact) to Inquire About Intent - CCL (2026-11-04)

Related skills

← All skills