afrexai-performance-review（Afrexai 绩效评估）

Performance Review Engine

Your AI-powered performance management system. Write reviews that develop people, not just evaluate them. From self-assessments to 360° feedback to calibration — complete frameworks for every review cycle.

Quick Start

Tell your agent:

• "Write a performance review for [name] — they exceeded on delivery but need to improve communication"
• "Help me write my self-assessment for H2 2025"
• "Run a 360° feedback collection for my team of 6"
• "Prepare calibration notes for my 4 direct reports"

1. Review Cycle Setup

Cycle Configuration Template

cycle:
  name: "H2 2025 Performance Review"
  period: "2025-07-01 to 2025-12-31"
  type: annual | semi-annual | quarterly
  timeline:
    self_assessment_due: "2026-01-10"
    peer_feedback_due: "2026-01-17"
    manager_draft_due: "2026-01-24"
    calibration_session: "2026-01-28"
    delivery_window: "2026-01-29 to 2026-02-07"
  participants:
    - name: ""
      role: ""
      level: ""
      tenure_months: 0
      previous_rating: ""
      peer_reviewers: []
      skip_level_reviewer: ""
  rating_scale:
    1: "Does Not Meet Expectations"
    2: "Partially Meets Expectations"
    3: "Meets Expectations"
    4: "Exceeds Expectations"
    5: "Significantly Exceeds Expectations"
  competencies:
    - name: "Delivery & Execution"
      weight: 30
    - name: "Technical/Functional Expertise"
      weight: 25
    - name: "Communication & Collaboration"
      weight: 20
    - name: "Leadership & Influence"
      weight: 15
    - name: "Growth & Development"
      weight: 10

Rating Distribution Guidelines

Forced distribution warning: These are guidelines, not quotas. If a team genuinely has 80% high performers, the distribution should reflect reality. Forcing bell curves creates distrust.

2. Self-Assessment Framework

STAR-I Method (Situation → Task → Action → Result → Impact)

Guide employees to write self-assessments that actually demonstrate value:

### Achievement: [Title]

**Situation:** What was the context or challenge?
**Task:** What was your specific responsibility?
**Action:** What did you do? (Be specific — tools, approaches, decisions)
**Result:** What was the measurable outcome?
**Impact:** How did this affect the team/org/company beyond the immediate result?

**Competency alignment:** [Which competency does this demonstrate?]
**Evidence:** [Links, metrics, Slack messages, PRs, customer feedback]

Self-Assessment Prompts by Competency

Delivery & Execution:

• What were your top 3-5 deliverables this period?
• Which projects were on time/budget? Which weren't, and why?
• How did you handle blockers or scope changes?
• What did you ship that you're most proud of?

Technical/Functional Expertise:

• What new skills or knowledge did you develop?
• Where did you serve as the go-to expert?
• What technical decisions did you make and what was the outcome?
• How did you stay current in your field?

Communication & Collaboration:

• How did you contribute to team effectiveness?
• Give an example of resolving a disagreement productively
• How did you share knowledge with others?
• What cross-functional work did you do?

Leadership & Influence:

• How did you influence outcomes beyond your direct responsibilities?
• Did you mentor or develop others? How?
• What initiatives did you drive or champion?
• How did you contribute to team culture?

Growth & Development:

• What feedback did you receive and act on?
• What's your biggest area of growth this period?
• Where do you still want to improve?
• What are your goals for next period?

Self-Assessment Quality Checklist

• [ ] Includes 5-8 concrete achievements with metrics
• [ ] Uses STAR-I format (not just "I did X")
• [ ] Covers all competency areas, not just delivery
• [ ] Acknowledges at least 1-2 growth areas honestly
• [ ] References specific feedback received and actions taken
• [ ] Includes forward-looking goals
• [ ] Avoids vague language ("helped with," "was involved in")
• [ ] Links to evidence where possible
• [ ] Appropriate length (1-2 pages, not 10)
• [ ] Written in first person, professional but human tone

3. Manager Review Writing

The OBSERVE Framework

Structure every review around:

O — Outcomes delivered: What did they ship/achieve? Metrics and evidence. B — Behaviors demonstrated: HOW they worked, not just what they produced. S — Strengths to leverage: Their superpower — what should they do MORE of? E — Edges to develop: Growth areas framed as opportunities, not failures. R — Relationships & impact: How they affected team dynamics and culture. V — Vision forward: Clear expectations and development plan for next period. E — Evidence cited: Every claim backed by specific examples.

Writing Rules

1. Specific > Vague

- ❌ "Great job this quarter" - ✅ "Led the API migration affecting 12 services, completing 2 weeks ahead of schedule with zero customer-facing incidents"

1. Behavior > Trait

- ❌ "Is a natural leader" - ✅ "Organized weekly knowledge-sharing sessions that improved team velocity by 15% and reduced onboarding time for 3 new hires"

1. Pattern > Incident

- ❌ "Missed the Q3 deadline" - ✅ "Delivery timelines were missed on 3 of 5 projects, consistently by 1-2 weeks, suggesting estimation needs improvement"

1. Forward > Backward

- ❌ "Failed to communicate effectively" - ✅ "Strengthening stakeholder communication — specifically proactive status updates — would multiply the impact of their strong technical work"

1. Balanced always

- Even top performers need development feedback - Even struggling performers have strengths to acknowledge - Target ratio: 60% strengths / 40% development (adjust by rating)

Review Templates by Rating

Rating 5 — Significantly Exceeds

## Performance Review: [Name] — H2 2025
**Rating: Significantly Exceeds Expectations (5/5)**

### Summary
[Name] delivered exceptional results this period, consistently operating above their current level. Their impact extended well beyond their role, influencing [team/org/company] outcomes in measurable ways.

### Key Achievements
1. **[Achievement]** — [STAR-I summary with metrics]
2. **[Achievement]** — [STAR-I summary with metrics]
3. **[Achievement]** — [STAR-I summary with metrics]

### Competency Assessment
| Competency | Rating | Evidence |
|-----------|--------|----------|
| Delivery & Execution | 5 | [Specific examples] |
| Technical Expertise | [X] | [Specific examples] |
| Communication | [X] | [Specific examples] |
| Leadership | [X] | [Specific examples] |
| Growth | [X] | [Specific examples] |

### Strengths to Leverage
- [Strength 1] — this is a differentiator that should be amplified
- [Strength 2] — consider giving them a platform to share this more broadly

### Development Opportunities
Even at this exceptional level, continued growth in [area] would unlock [next-level impact]. Specifically:
- [Development area with actionable suggestion]
- [Stretch assignment or learning recommendation]

### Forward Look
[Name] is ready for [promotion/expanded scope/leadership opportunity]. Recommended next steps: [specific action].

Rating 3 — Meets Expectations

## Performance Review: [Name] — H2 2025
**Rating: Meets Expectations (3/5)**

### Summary
[Name] delivered solid, reliable work this period, meeting the expectations of their role. They are a dependable contributor who [key positive theme].

### Key Achievements
1. **[Achievement]** — [Evidence]
2. **[Achievement]** — [Evidence]
3. **[Achievement]** — [Evidence]

### Competency Assessment
[Same table format]

### Strengths
- [Strength 1 with evidence]
- [Strength 2 with evidence]

### Development Areas
To move from "meets" to "exceeds," [Name] should focus on:
1. **[Area]** — Currently [current state]. The gap is [specific gap]. To close it: [actionable steps].
2. **[Area]** — [Same structure]

### Forward Look
Goals for next period:
1. [Measurable goal tied to development area]
2. [Stretch goal that would demonstrate growth]
3. [Continuation goal building on strengths]

Rating 1-2 — Below Expectations

## Performance Review: [Name] — H2 2025
**Rating: [Partially Meets / Does Not Meet] Expectations ([1-2]/5)**

### Summary
[Name] struggled to meet expectations in key areas this period. While [acknowledge any positives], significant gaps in [areas] need to be addressed.

### Performance Gaps
1. **[Gap]** — Expected: [what was expected]. Actual: [what happened]. Impact: [business impact]. Examples: [2-3 specific instances].
2. **[Gap]** — [Same structure]

### What Was Done Well
- [Genuine positive — never skip this section]

### Context Considered
- [Any mitigating factors: reorg, unclear expectations, personal circumstances]
- [Whether support/coaching was provided and when]

### Improvement Plan
| Area | Current State | Target State | Actions | Timeline | Support Needed |
|------|--------------|-------------|---------|----------|----------------|
| [Gap 1] | [Specific] | [Specific] | [Steps] | [Date] | [Resources] |
| [Gap 2] | [Specific] | [Specific] | [Steps] | [Date] | [Resources] |

### Consequences
If improvement to [specific measurable standard] is not demonstrated by [date]:
- [Next step: PIP / role change / separation]

### Check-in Schedule
- Weekly 1:1s focused on [areas]
- 30-day checkpoint: [date]
- 60-day checkpoint: [date]
- Final assessment: [date]

4. 360° Feedback System

Peer Feedback Request Template

Hi [Peer Name],

You're invited to provide feedback on [Employee Name] for our [H2 2025] review cycle.

Please share your observations (10-15 min, ~200-400 words total):

1. **What does [Name] do well?** (Think: specific projects, behaviors, impact on you/the team)
2. **What could [Name] improve?** (Think: what would make them even more effective?)
3. **How would you describe working with [Name]?** (Collaboration style, communication, reliability)
4. **One thing [Name] should keep doing:** ___
5. **One thing [Name] should start or do more of:** ___

Your feedback will be anonymized and synthesized — [Name] will not see your individual responses verbatim.

Due by: [Date]

Feedback Synthesis Method

When combining multiple peer reviews:

1. Identify themes — What do 2+ people mention? Those are patterns, not noise.
2. Weight by proximity — Feedback from close collaborators > occasional contacts.
3. Separate fact from feeling — "Missed 3 deadlines" is fact. "Seems disengaged" is perception (still valuable, but frame differently).
4. Preserve outlier insights — If one person noticed something unique, it may still be valuable. Include as "additionally noted."

Synthesis Template

### 360° Feedback Summary for [Name]

**Respondents:** [N] peers, [N] cross-functional, [N] skip-level

**Consistent Strengths (mentioned by 2+ reviewers):**
- [Theme] — "[Representative quote]" (paraphrased from [N] responses)
- [Theme] — "[Representative quote]"

**Consistent Development Areas:**
- [Theme] — "[Representative quote]"
- [Theme] — "[Representative quote]"

**Notable Individual Observations:**
- [Unique insight worth including]

**Overall Sentiment:** [Positive / Mixed / Concerning]
**Collaboration Rating (aggregated):** [Strong / Solid / Needs Improvement]

5. Calibration Session

Pre-Calibration Prep

For each direct report, prepare:

calibration_card:
  name: ""
  current_level: ""
  tenure: ""
  previous_rating: ""
  proposed_rating: ""
  rating_justification: "" # 2-3 sentences max
  top_achievement: ""
  biggest_gap: ""
  promotion_candidate: yes | no | not_yet
  flight_risk: low | medium | high
  key_question: "" # What you want the calibration group to weigh in on

Calibration Discussion Framework

Round 1 — Present (2 min per person)

• Manager presents: proposed rating, top achievement, biggest gap
• No debate yet — just laying out the landscape

Round 2 — Calibrate (5 min per person where needed)

• Focus on: rating 4s and 5s (are they truly exceptional?), rating 1s and 2s (is this fair?), any rating that changed from last cycle
• Ask: "Would this person get the same rating on another team?"
• Ask: "Is this rating consistent with [comparable person]?"

Round 3 — Decide

• Finalize ratings
• Flag anyone who needs skip-level review
• Identify promotion candidates
• Identify flight risks needing retention action

Calibration Bias Checklist

Before finalizing, check for:

• [ ] Recency bias — Are you over-weighting the last month vs. the full period?
• [ ] Halo/horns effect — Is one great/bad thing coloring the entire review?
• [ ] Similarity bias — Are you rating people like you higher?
• [ ] Central tendency — Are you avoiding extreme ratings when they're warranted?
• [ ] Leniency/strictness — Is your distribution shifted vs. the org?
• [ ] Attribution error — Are you blaming the person for systemic issues?
• [ ] Contrast effect — Are you comparing to the previous person reviewed rather than the standard?

6. Review Delivery Conversation

Conversation Structure (45-60 min)

Opening (5 min)

• Set the tone: "This is a two-way conversation, not a verdict"
• Share the rating upfront — don't make them wait

Achievements (10 min)

• Walk through top 3-5 achievements
• Let them add context or achievements you missed
• Be genuinely appreciative — this isn't just preamble to criticism

Development (15 min)

• Present 1-2 development areas (not 10)
• Use the pattern: "I've observed [specific behavior] in [specific situations]. The impact was [what happened]. What I'd love to see is [desired behavior]."
• Ask: "Does this resonate? What's your perspective?"
• Listen. Actually listen.

360° Themes (5 min)

• Share synthesized peer feedback
• Highlight: "Your colleagues really value [X]"
• Development: "A theme that came up was [Y] — thoughts?"

Goals & Development Plan (15 min)

• Co-create 3-5 goals for next period
• At least 1 development goal, not just delivery goals
• Identify specific actions, resources, support needed
• Agree on check-in cadence

Close (5 min)

• Summarize key takeaways
• Ask: "What do you need from me to be successful?"
• End on forward-looking, supportive note

Difficult Conversation Scripts

For underperformers: "I want to be direct with you because I respect you and your potential here. Your performance this period was below what we need in [specific area]. Here's what I've observed... I want to work with you on a plan to get back on track. Are you willing to commit to that?"

For strong performers who didn't get promoted: "Your work this period was excellent — [specific examples]. The reason you're rated [X] rather than promoted is [specific gap]. Here's what I think it would take: [concrete steps]. I'm committed to supporting you in getting there."

For someone who disagrees with their rating: "I hear you, and I want to understand your perspective. Can you walk me through the specific areas where you see it differently? ... I appreciate you sharing that. Here's how I weighed [factors]. [Either: Let me take this back and reconsider / I understand the disagreement, but here's why the rating stands]."

7. Development Planning

Development Plan Template

development_plan:
  employee: ""
  manager: ""
  period: "H1 2026"
  review_date: ""
  
  strengths_to_leverage:
    - strength: ""
      leverage_action: "" # How to use this more
      
  development_areas:
    - area: ""
      current_state: ""
      target_state: ""
      actions:
        - type: "on_the_job" # 70% of development
          description: ""
          timeline: ""
        - type: "learning" # 20% — coaching, mentoring, peer learning
          description: ""
          timeline: ""
        - type: "formal" # 10% — courses, certifications, conferences
          description: ""
          timeline: ""
      success_metrics: ""
      check_in_dates: []
      
  career_goals:
    short_term: "" # 6-12 months
    medium_term: "" # 1-3 years
    long_term: "" # 3-5 years
    
  support_needed:
    from_manager: ""
    from_org: ""
    budget_required: ""

The 70-20-10 Development Mix

Common mistake: Over-indexing on formal learning (sending someone to a course) when on-the-job stretch would be 5x more effective.

8. Continuous Feedback (Between Reviews)

1:1 Performance Check-in Template (Monthly)

## Monthly Check-in: [Name] — [Month Year]

### Progress on Goals
| Goal | Status | Notes |
|------|--------|-------|
| [Goal 1] | 🟢 On track / 🟡 At risk / 🔴 Off track | [Brief update] |

### Recent Wins
- [What went well this month]

### Challenges
- [What's been difficult]

### Feedback Exchange
- **Manager → Employee:** [One specific piece of feedback]
- **Employee → Manager:** [Ask: "What can I do differently to support you?"]

### Action Items
- [ ] [Action] — Owner: [who] — By: [date]

### Overall Pulse: 😊 Great / 😐 Fine / 😟 Struggling

Real-Time Feedback Formula (SBI)

Situation: "In yesterday's client presentation..." Behavior: "...you handled the pricing objection by reframing around ROI rather than discounting..." Impact: "...which kept us at full price and the client visibly shifted from skeptical to interested."

Deliver within 48 hours. Positive feedback publicly (if they're comfortable). Constructive feedback privately. Always.

9. Scoring & Analytics

Individual Performance Score (0-100)

Score = Σ (competency_rating × competency_weight) × 20

Example:
Delivery (4/5 × 30%) + Technical (3/5 × 25%) + Communication (4/5 × 20%) 
+ Leadership (3/5 × 15%) + Growth (4/5 × 10%)
= (1.20 + 0.75 + 0.80 + 0.45 + 0.40) = 3.60 / 5 = 72/100

Team Health Dashboard

Track quarterly:

## Team Performance Dashboard — Q4 2025

**Team size:** [N]
**Rating distribution:** ⭐5: [N] | ⭐4: [N] | ⭐3: [N] | ⭐2: [N] | ⭐1: [N]
**Average score:** [X]/100
**vs. last period:** [↑/↓ X points]

**Promotion candidates:** [Names]
**Flight risks:** [Names + risk level]
**PIP/coaching:** [Names]

**Top team strengths:** [Competencies scoring highest]
**Team gaps:** [Competencies scoring lowest]
**Development budget used:** [X]% of [Y] allocated

**Engagement signals:**
- Voluntary turnover: [X]%
- Internal mobility: [X] transfers/promotions
- 1:1 completion rate: [X]%
- Goal completion rate: [X]%

10. Edge Cases & Advanced Scenarios

New Hire (< 6 months)

• Evaluate against onboarding milestones, not full role expectations
• Weight learning speed and cultural integration higher
• Compare to "expected ramp" not to tenured peers
• Rating floor of 3 unless genuine performance issues (distinguish slow ramp from bad fit)

Role Change Mid-Cycle

• Split the review: first half in old role, second half in new
• Weight the new role performance more heavily (it's the forward-looking signal)
• Acknowledge the transition tax — expect a temporary dip

Remote/Hybrid Considerations

• Evaluate output and impact, not visibility or hours
• Seek feedback from async collaborators, not just people in the office
• Watch for proximity bias — don't rate in-office people higher by default

High Performer Wanting to Leave

• Have the conversation: "I value you and want to understand what would make you want to stay"
• Don't inflate the rating as retention — it sets a precedent
• Document the conversation and retention actions taken

Inherited Team Member

• Get context from previous manager (ask for their calibration card)
• Be transparent: "I'm still building my understanding of your work"
• Lean more on peer feedback and objective metrics
• Don't default to "meets" because you don't know — do the research

Manager Reviewing Someone They Don't Like

• Stick to observable behaviors and measurable outcomes
• Have a peer manager gut-check your review for bias
• Ask yourself: "If my favorite team member did exactly this, what would I rate them?"

11. Legal & Compliance Notes

Documentation rules:

• Keep all review documents for minimum 3 years (7 in regulated industries)
• Feedback must reference specific, observable behaviors — not personality traits
• Never reference protected characteristics (age, gender, disability, etc.)
• PIP documentation should be reviewed by HR/legal before delivery
• Employee should sign acknowledging receipt (not agreement)

Phrases to avoid:

• "Cultural fit" (can mask bias) → Use "collaboration effectiveness"
• "Aggressive" (gendered connotation) → Use "assertive" or "direct"
• "Young/energetic" → Use specific behaviors
• "Not a team player" → Cite specific collaboration gaps with examples

Commands Reference