The Uncomfortable Truth About Performance Reviews

Let me start with a confession: I used to play favorites in performance reviews.

Not intentionally, of course. I would never consciously give better reviews to people I liked or had more visibility with. But looking back at my reviews from three years ago, the pattern was undeniable:

• Recency bias: The engineers I worked with in November got more detailed, glowing reviews than those whose biggest wins happened in February
• Visibility bias: The vocal engineer who Slacked me daily got more recognition than the quiet one who consistently delivered
• Similarity bias: I wrote longer, more thoughtful reviews for people who worked like I did
• Halo effect: One impressive project colored my entire evaluation of that person
• Inconsistency: My 10th review of the day was noticeably shorter and less thoughtful than my 1st

I wasn't a bad manager. I was just... human.

And that's the problem.

The Hidden Cost of Biased Reviews

Performance reviews aren't just paperwork. They directly impact:

• Compensation - Raises and bonuses
• Promotions - Who moves up the ladder
• Opportunities - Who gets the high-visibility projects
• Development - What training and mentorship people receive
• Retention - Whether your best people stay or leave

When reviews are inconsistent or biased, even unintentionally, you:

• Lose talented people who feel undervalued
• Promote the wrong people for the wrong reasons
• Create a culture where visibility matters more than results
• Spend political capital dealing with appeals and disputes
• Lose your team's trust

The stakes are too high for "I'm only human" to be an acceptable answer.

The Traditional Solution (That Doesn't Scale)

The conventional wisdom for reducing bias is:

1. Keep detailed notes all year
2. Use structured rubrics
3. Gather 360-degree feedback
4. Calibrate with peer managers
5. Review for consistency

These are all good practices. I do all of them.

But here's the reality when you manage a team of 10+ people:

Each performance review takes 4-6 hours:

• 1 hour reviewing notes and data
• 1 hour analyzing against objectives
• 2 hours writing the review
• 1 hour editing and refining
• 1 hour in the review meeting

That's 40-60 hours of focused work for a 10-person team. And by hour 40, you're exhausted. Your 10th review isn't getting the same thoughtfulness as your 1st.

You end up cutting corners. You reuse phrases. You rush. Quality suffers.

And unconscious bias creeps back in.

The AI Solution: Scale Quality, Eliminate Bias

This year, I tried something different.

I built a system using Claude AI that helps me write consistent, thorough, unbiased performance reviews at scale—without losing the human judgment that makes them valuable.

The result?

• Time per review: 90 minutes (down from 4-6 hours)
• Consistency: 100% (same structure and depth for everyone)
• Bias reduction: Measurable improvement (more on this below)
• Employee satisfaction: Noticeably higher

The System: Structured Input, Consistent Output

The Problem I Solved

Traditional performance review writing has two fundamental issues:

1. Unstructured input - Managers work from scattered notes, memories, and impressions
2. Inconsistent output - Each review is written from scratch with different levels of detail and thought

My AI-assisted system fixes both.

The Process

Step 1: Structured Data Collection (I do this)

I maintain a standardized template for each employee that captures:

markdown

## Employee Metadata
- Name, role, department
- Review period
- Performance level (I decide this)
- Lead/behavior level (I decide this)

## Business Objectives (3-5 per person)
- What was the goal? (from annual planning)
- How were they supposed to achieve it?
- What actually happened? (my assessment)
- Achievement level: Not Met / Partially Met / Fully Met / Exceeded

## Development Objectives
- What skills/certifications were they pursuing?
- What progress did they make?
- What training did they complete?

## Company Assessment Criteria (uploaded once)
- SMART objectives framework (PDF)
- WHAT assessment guide (PDF)
- HOW behavioral framework (PDF)
- Performance level definitions (PDF)

## Employee's Self-Submitted Review (from HR system)
- Their self-assessment (PDF)
- Their perspective on the year

Key insight: I'm not asking AI to evaluate the employee. I'm providing my evaluation in structured form.

Step 2: AI Processing (Claude does this)

I feed this structured data to Claude with a detailed prompt that says:

"You are helping me write a performance review. I have already decided the performance levels and provided all my assessments. Your job is to:

1. Write consistent, professional prose based on my inputs
2. Ensure each section follows company framework (SMART, WHAT/HOW)
3. Maintain the same depth and quality for every employee
4. Use specific examples I provided
5. Keep tone constructive and forward-looking
6. Follow the exact template structure

Do NOT change my assessments. Do NOT add subjective opinions. Transform my structured data into polished review prose."

Step 3: Review and Refine (I do this)

I review the AI-generated draft and:

• Verify accuracy of all facts
• Add any missing context or nuance
• Adjust tone where needed
• Ensure it reflects my actual feedback

This takes 30-45 minutes per person—a fraction of the original time.

Step 4: Employee Meeting (I do this)

The AI never replaces the human conversation. I still:

• Meet with each employee personally
• Discuss the review face-to-face
• Answer questions and provide context
• Collaboratively set next year's goals

In Part 2, I will explain how implementation works - and why it actually improves quality rather than diminishing it.

The Step-by-Step Implementation Guide

Want to try this yourself? Here's exactly how to set it up:

Phase 1: Create Your Template (1-2 hours)

Below is the template that I created.

Dont give EMPLOYEE_IMPACT_LEVEL and EMPLOYEE_LEAD_LEVEL ratings and you can ask AI that based on the review what RATINGS an employee deserves.

You will be surprised that the result will be without biased and exactly what you thought without biasedness.

# Performance Review 2025 - **{{ EMPLOYEE_NAME }}**

## Employee Information
**Instructions:** Fill in the basic information below. This will auto-populate throughout the review.

- **EMPLOYEE_NAME**: EMPLOYEE_NAME
- **POSITION**: EMPLOYEE_POSITION
- **DEPARTMENT**: EMPLOYEE_DEPARTMENT
- **REVIEW_PERIOD**: [January 2025 - December 2025]
- **EMPLOYEE_IMPACT_LEVEL**: 
- **EMPLOYEE_LEAD_LEVEL**: 
- **PERFORMANCE_REVIEW**: 2025
- **COMPANY_SMART_PDF**: 01_SMART.pdf
- **COMPANY_WHAT_PDF**: 02_WHAT.pdf
- **COMPANY_HOW_PDF**: 03_HOW.pdf
- **COMPANY_PLD_PDF**: 04_PLD.pdf
- **EMPLOYEE_PERFORMANCE_REVIEW_PDF**: 05_EMPLOYEE_NAME_PR_2025.pdf

---
## Business Objectives

**Instructions:** For each objective, describe WHAT was to be achieved, HOW it was to be accomplished, and provide your evaluation comments with specific examples and outcomes.

### Objectives

**Reference Documents:**
- Review **{{ EMPLOYEE_PERFORMANCE_REVIEW_PDF }}** for objective assessment criteria under **Annual Business Objectives** inside the PDF

**Manager's Evaluation Comments:**
- Achievement Level: **{{ EMPLOYEE_IMPACT_LEVEL }}**

---
## Development Objectives

**Instructions:** Document professional development goals and progress.

### Objectives
**Reference Documents:**
- Review **{{ EMPLOYEE_PERFORMANCE_REVIEW_PDF }}** for objective assessment criteria under **Annual Development Objective** inside the PDF

**Manager's Evaluation Comments:**
- Achievement Level: **{{ EMPLOYEE_LEAD_LEVEL }}**

---
## Impact Assessment (WHAT)

**Instructions for Manager:** 
This section assesses the IMPACT the employee had on business objectives throughout the year.

**Reference Documents:**
- Review **{{ COMPANY_SMART_PDF }}** for objective assessment criteria
- Review **{{ COMPANY_WHAT_PDF }}** for guiding questions
- Review **{{ COMPANY_PLD_PDF }}** for level definitions
- Review **{{ EMPLOYEE_PERFORMANCE_REVIEW_PDF }}**

**Selected Impact Level:** **{{ EMPLOYEE_IMPACT_LEVEL }}**

**Impact Level Definition:**
[The system will auto-fill based on selection:]
- **Fully Meets Expectations:** Meets all expectations even in difficult context. Requires little to no additional direction to achieve the core goals of the role.
- **Mostly Meets Expectations:** Mostly meets standard objectives in a stable context. Achieves only part of core goals for the role.
- **Exceeds Expectations:** Over performs, even when objectives are stretched. Delivers beyond the scope of the role and succeeds even on additional missions or transverse actions.
- **Below Expectations:** Does not fulfill a majority of standard objectives in a stable context.

### Manager's Impact Assessment (WHAT)
**Instructions:** Write 5 professional sentences assessing **{{ EMPLOYEE_NAME }}**'s yearly impact. Consider:
- What outcomes did they achieve across all objectives?
- What difficulties/enablers did they face?
- Did they reach defined targets and milestones?
- What are you most proud of in their achievements?
- How could they have had more impact?

**Your Assessment:**
[Write your 5-sentence assessment here - this will be AI-generated based on all inputs above]

---
## Lead@Company Implementation (HOW)

**Instructions for Manager:**
This section assesses HOW the employee worked - their behaviors, collaboration, and leadership approach throughout the year.

**Reference Documents:**
- Review **{{ COMPANY_HOW_PDF }}** for the three dimensions: THINK BIG, MAKE IT HAPPEN, TOGETHER
- Review **{{ COMPANY_PLD_PDF }}** for level definitions
- Review **{{ EMPLOYEE_PERFORMANCE_REVIEW_PDF }}**

**Selected Lead Level:** **{{ EMPLOYEE_LEAD_LEVEL }}**

**Lead Level Definition:**
[The system will auto-fill based on selection:]
- **Fully Meets Expectations:** Demonstrates the L@T behaviours consistently and actively seeks to develop their behaviours.
- **Mostly Meets Expectations:** Mostly demonstrates the L@T behaviours and engages in development actions to improve them.
- **Exceeds Expectations:** Acts as role model in L@T, seeks to develop themselves and others, and is involved in wider action plans.
- **Below Expectations:** Does not demonstrate the L@T behaviours and does not actively seek to develop themselves.

### Manager's Lead@Company Assessment (HOW)
**Instructions:** Write 4 professional sentences assessing **{{ EMPLOYEE_NAME }}**'s behaviors and approach. Consider:

**THINK BIG:**
- How did they communicate purpose and vision?
- What innovative ideas did they propose?
- How did they handle complexity?

**MAKE IT HAPPEN:**
- How did they build trust and deliver results?
- How did they measure progress?
- What risks did they take?

**TOGETHER:**
- How did they contribute to team performance?
- How did they cooperate with others?
- How did they handle difficult situations?

**Your Assessment:**
[Write your 4-sentence assessment here - this will be AI-generated based on all inputs above]

---
## Workload Assessment

**Instructions for Manager:** Assess whether the workload was appropriate for achieving the defined objectives.

**Question:** Was **{{ EMPLOYEE_NAME }}**'s workload compatible with achieving the defined objectives?

**Your Response:** [Select: Yes]

**Manager's Workload Evaluation:**
**Instructions:** Write 4 professional sentences addressing:
- Was the workload appropriate and balanced?
- Did they have adequate resources, time, and support?
- Were there any environmental factors (organizational changes, market conditions) that affected performance?
- How did they manage competing priorities?

**Your Assessment:**
[Write your 4-sentence assessment here - this will be AI-generated]

---
## Overall Evaluation

### Performance Summary
- **Impact (WHAT):** **{{ EMPLOYEE_IMPACT_LEVEL }}**
- **Lead (HOW):** **{{ EMPLOYEE_LEAD_LEVEL }}**

### Performance Level Definitions Applied

**Impact Level - {{ EMPLOYEE_IMPACT_LEVEL }}:**
- Manager Assessment - **{{ EMPLOYEE_IMPACT_LEVEL }}**.

**Lead Level - {{ EMPLOYEE_LEAD_LEVEL }}:**
- Manager Assessment - **{{ EMPLOYEE_LEAD_LEVEL }}**.

---
### Employee Self-Assessment Comment 

**Instructions for Employee:** Reflect on your year - achievements, challenges, growth, and future aspirations.

[Employee to complete this section:]

Review **{{ EMPLOYEE_PERFORMANCE_REVIEW_PDF }}** for Employee comment under **Employee Overall Input** inside the PDF

---
### Manager's Overall Evaluation

**Instructions for Manager:** Write 6 comprehensive sentences that synthesize the entire review. Structure your evaluation to:

1. **Start with strengths and achievements** - Lead with what they did well, using specific examples from objectives above
2. **Acknowledge their impact** - Highlight their most significant contributions to the team/organization
3. **Address development areas constructively** - Frame areas for growth as opportunities, not criticisms
4. **Provide concrete examples** - Reference specific situations, projects, or behaviors mentioned above
5. **Connect to feedback already given** - Ensure nothing here is a surprise; it should reflect ongoing conversations
6. **End with forward-looking focus** - Set the tone for next year with clear areas of focus or growth opportunities

**Use natural, authentic language** - Avoid corporate jargon. Write as you would speak to them directly.
**Balance recognition with development** - Make them feel valued while motivating continued growth.

**Your Overall Evaluation:**
[Write your 6-sentence overall evaluation here - this will be AI-generated based on all sections above]

Phase 2: Gather Company Documents (30 minutes)

Collect and digitize:

• Company performance framework (PDF)
• Evaluation criteria (PDF)
• Behavioral competency model (PDF)
• Performance level definitions (PDF)

These become your "source of truth" that AI references.

Phase 3: Create Your Prompt (1 hour)

Write a detailed prompt for your AI that includes previous prompt and company and employee reference documents.

Run this with help of below prompt one by one or together for all employees

You are helping me write performance reviews for my team.

CONTEXT:
- I manage [number] engineers in [department]
- Our company uses [framework name] for evaluations
- I have already decided all ratings and assessments

YOUR ROLE:
Transform my structured assessment data into polished 
performance review prose.

REQUIREMENTS:
1. Follow the exact template structure I provide
2. Use company framework from the PDFs I upload
3. Write in professional, constructive tone
4. Include all specific examples I provide
5. Maintain consistent depth (1500-1800 words)
6. Do NOT change my ratings or assessments
7. Do NOT add subjective opinions

OUTPUT:
A complete performance review ready for employee 
discussion, following our template exactly.

The Results: My First Year

Here's what happened when I implemented this system for my 12-person team:

Time Savings

• Before: 30+ hours writing reviews
• After: 5+ hours (total for all 12 people)
• Saved: 20 hours (nearly a 3 full work days!)

Quality Improvements

• More specific examples in every review
• Consistent depth across all team members
• Better alignment with company framework
• Fewer appeals or questions about ratings

That last question is why I'm writing this post.

The Ethics of AI-Assisted Reviews

Let me address the elephant in the room: Is it ethical to use AI for something as important as performance reviews?

I believe the answer is yes—if done right.

Here's my ethical framework:

1. Transparency

I tell my team I use AI to help structure reviews. Most appreciate it—they see it as me investing in quality feedback.

2. Human Accountability

I am 100% responsible for every word in the review. AI is a tool, not a decision-maker.

3. Bias Reduction

Using AI has measurably reduced my unconscious bias. That's more ethical than pretending humans are perfectly fair.

4. Consistency as Fairness

Giving everyone the same level of attention and structure is fairer than favoring whoever I happen to remember most clearly.

5. Time for What Matters

Saving 30+ hours lets me spend more time:

• Having better 1:1 conversations
• Mentoring and coaching
• Observing and taking notes all year
• Supporting career development

Final Thoughts

If you are writing biased reviews and didn't even know it.

With help of AI you can write fairer, more consistent reviews in a fraction of the time—and your team will appreciate the improvement.

AI didn't replace manager’s judgment. It freed manager to apply his judgment more consistently and fairly to everyone on the team.

AI didn't make reviews impersonal. It gave manager more time for the personal conversations that matter.

AI didn't eliminate my responsibility. It helps manager to fulfill that responsibility better.

The technology is here. The question is: Will you use it to become a better manager?

Society punishes material theft but ignores identity theft of the soul.

Material abundance may someday end object theft. Personality theft? Never.

When you steal personalities, you change yourself—subtly, daily, relentlessly. Over time, you transform so completely you no longer recognize who you are.

Don’t steal specially personalities. Stay authentic. Be unapologetically yourself.

An actor once said: “I can do everything with ease on the stage, whereas in real life I feel too big and clumsy.”

Don’t live as a character. Live as yourself.

The Problem: Information Overload

Market research generates dense reports. My Claude research query on in-flight entertainment systems returned 15 pages of structured analysis covering oligopolies, financial metrics, competitive matrices, and regulatory frameworks. Valuable insights buried in walls of text.

Reading time: 2+ hours. Attention span: 15 minutes.

The Solution: Automate the Presentation Layer

Instead of manually building a dashboard, I meta-programmed the process:

Step 1: Generate Research (11m 26s)

Input: Market structure analysis prompt for IFE systems
Mode: Claude Web + Research
Output: Claude.pdf (15 pages, investor-grade dossier)

Key requirement: The prompt enforced strict sourcing rules—only primary filings (10-K, 20-F), regulatory documents, and verified tier-one research. Every metric traceable or tagged "single-source (low confidence)."

The prompt is like below. Please change the sector name to your sector name.

My sector name - In-flight entertainment (IFE) systems for commercial aviation

## ROLE
Market-structure analyst for long-term investors. Produce an investor-grade, qualitative-first dossier.
Facts must be traceable and cross-verified. If evidence is weak, say so.

## INPUTS
sector = YOUR_SECTOR_NAME

## OBJECTIVE
Understand 
- how the sector is structured
- who the true players are, and 
- how competition works.

Deliver a 1-page executive overview + a full structured analysis with compact tables.

## ENFORCEMENT RULES (NO GUESSING)
Source bounds: 
- use only A–D. 
- If none available → write “Insufficient evidence.”

A) Primary filings/IR (10-K/20-F/URD/S-1, audited IFRS/GAAP, investor days, transcripts).
B) Regulators/statistics (industry regulators, national stats, IMF/OECD, BIS, etc.).
C) Official price lists, methodologies, technical specs, product catalogs, rulebooks.
D) Tier-one journals/consultancies with disclosed primary data; major data vendors with methodology.

### Verification
Any figure must be checked against ≥2 independent A–C sources → else tag “single-source (low confidence).”

### Uncertainty
If unsure → write “Unknown.” Do not estimate.

### Conflicts
If sources differ >5% → show both, define scope/period, and explain likely cause.

### Recency
Prefer ≤24 months; older = “Legacy.”

### Citations
Bracketed footnotes [1], [2], mapping to a Source List (≤15 items).

## UNIT OF COMPETITION (MANDATORY) :
For the given sector, automatically identify:

1. Lanes = the 5 universal types of competition
(Only use those that exist in the sector.)
- Upstream Inputs / Extraction
- Core Production / Infrastructure / Operations
- Distribution / Access / Channels
- Data / IP / Standards / Certification
- Services / Wrap-around Support

2. Customer groups = the 5 universal buyer types
(Only use those that exist in the sector.)
- End consumers
- Enterprise/professional buyers
- Intermediaries/distributors
- Regulators/standards bodies
- Complementors/ecosystem partners

3. Define true peers per lane
- Do not compare players across unrelated lanes.

## OUTPUTS
A) Executive Overview (≤300 words)
- Sector structure, lanes, main players, differentiation, constraints, catalysts.

B) Player Atlas (qualitative)
- For every significant player in each lane:
- Legal/brand; ownership; geography in North America, Europe, Asia-Pacific
- Vertical scope (upstream → core → distribution → data/IP → services).
- Products/offerings.
- Customers served.
- Why customers choose them (price, quality, reliability, service, design, IP, network, availability, latency, brand, regulation, etc.).
- Pricing model.
- Market-design specifics (industry-dependent).
- Competitive stance (local monopoly, fragmented, oligopoly, global scale, niche).
- Moats: network effects, IP, scale, regulation, switching costs, ecosystem lock-in.
- Vulnerabilities: regulation, commoditization, disruption, outages, supply risk.

C) Peer Sheet — Normalized KPIs (per lane)
- Revenue/EBIT mix.
- Margin structure (EBITDA %, FCF conversion).
- ROCE/ROIC.
- Capex intensity.
- Pricing power indicators.
- Volume metrics relevant to the sector.
- Productivity/efficiency metrics.
- 10-year through-cycle resilience.

D) Moats & Switching Analysis
- Lane-by-lane qualitative assessment.

E) Customer-Choice Narratives
- For each customer group: why they choose one supplier over another.

F) Regime, Regulation, & Catalysts
- Regulatory constraints.
- Structural changes.
- M&A patterns (10-year).
- Technology shifts.
- Macro drivers and risks.

G) Watchlist (5–8 items)
Key KPIs, catalysts, and risks to monitor.

## COMPARATIVE MATRICES (MANDATORY)
1) Players × Customer-Choice Drivers
(price, quality, service, network, distribution, IP, regulation, reliability, etc.)

2) Players × Moats/Frictions
(network effects, IP, scale economics, regulatory entrenchment, switching costs)

3) Lane-based cluster map
Shows true peers vs adjacent players.

## FORMAT
Do the following:
- 1-page Executive Overview
- Full Player Atlas & Peer Sheet
- Two matrices + short narratives
- Appendices with
  - Assumptions
  - Methods, and
  - CSV-ready tables

## BEGIN
Do the following:
- Identify the lanes (from the universal categories).
- Map players into lanes; define true peers.
- Build Player Atlas.
- Populate Peer Sheet using evidence rules.
- Write customer-choice narratives + moat analysis.
- Add regime, catalysts, and watchlist.
- Deliver all sections.

Step 2: Generate Instructions (5 minutes)

Asked Claude to write deployment instructions for building a website from the PDF. The generated instructions.md specified:

• Tech stack: Next.js 14 + TypeScript + Tailwind + shadcn/ui
• Architecture: 6 page routes with tabbed interfaces
• Component specs: DataTable with color coding, PlayerCard layouts, MatrixView visualizations
• Design system: Professional BI aesthetic with semantic color palette (green=strong, yellow=moderate, red=weak)
• Data modeling: TypeScript interfaces for players, metrics, and competitive matrices

Step 3: Build the Site (30-40 minutes)

Dropped instructions.md into a new Cursor/Claude session. The AI generated:

• Responsive Next.js app with 6 main sections
• Interactive data tables with sorting and color-coded cells
• Tabbed player profiles (Panasonic, Thales, Viasat, Starlink, etc.)
• Competitive heat maps showing moats and customer drivers
• M&A timeline visualization
• KPI watchlist dashboard

Zero manual coding. Just iterative refinements through prompts.

Cursor Project looks like this

Technical Highlights

Color-Coded Intelligence

Tables use semantic colors automatically:

• ● (strong) → bg-green-100 text-green-800
• ◐ (moderate) → bg-yellow-100 text-yellow-800
• ○ (weak) → bg-gray-100 text-gray-600

Data Fidelity

The site preserves research rigor:

• "Not Disclosed" for missing data (never blank cells)
• Asterisks for single-source estimates
• Footnotes linking to verification sources
• Currency timestamps ("As of Q4 2024")

Mobile-First Tables

Complex financial tables adapt responsively:

• Desktop: Full matrix view
• Tablet: Horizontal scroll with sticky columns
• Mobile: Card-based layout

Website Page-1

Website Page-2

Why This Matters

Traditional workflow:

Workflow Comparison

Results

The deployed site includes:

• Executive overview with 6 key stat cards
• Player atlas with detailed profiles for 8+ companies
• Comparative matrices (Players × Moats, Players × Customer Drivers)
• Interactive M&A timeline (2015-2025)
• Regulatory framework documentation
• 8-metric watchlist dashboard

Run locally:

bash

npm install && npm run dev

Universal Applications

This approach works across multiple domains:

Business Intelligence:

• Competitive intelligence analysis (this example)
• Product comparison platforms
• Financial analysis dashboards

Knowledge Management:

• Technical documentation portals
• Educational course materials
• Internal wiki systems

Adapt to Any Industry

The research prompt is sector-agnostic. Simply modify the input:

sector = [Your Industry]

Examples:

• Cloud infrastructure providers
• SaaS security tools
• Industrial automation
• Pharmaceutical manufacturing
• Renewable energy storage

The instruction template automatically adapts to your domain, generating appropriate data structures, visualizations, and competitive frameworks.

The future of technical work transcends code volume. It demands exceptional specification writing, intuitive information architecture, and the ability to distill complex data into clear visual narratives.

More on SDD(Spec Driven Design) in future blogs