Overview

A manager dreads writing performance reviews. They have 10 to write. They ask AI: "Draft a performance review for someone who had a solid year." AI produces: "John demonstrated strong technical skills and collaborative approach. He delivered projects on time and contributed positively to team dynamics. He shows promise and should continue to develop his leadership capabilities."

This review is technically fine. It's also useless. It says nothing specific. John doesn't learn anything. HR doesn't learn anything. It's a template with his name plugged in.

The real review should say: "John shipped three features on schedule. The Q3 feature was complex and he asked good questions before starting, which prevented rework. He struggles with documentation, shipped three features with incomplete docs, and we talked about this. He's ready for a tech lead role; his next steps are mentoring a junior engineer and leading the architecture for the Q4 project."

This review is specific. It's actionable. It tells John where he excels and where he needs to grow. It forms a basis for his development.

This lesson teaches you how to use AI for reviews without producing generic praise. The key: you provide the specifics. AI helps you organize and articulate them well. The difference between a review that matters and a review that goes in a file and is forgotten is specificity.

Why This Matters for HR Professionals

Performance reviews affect people's lives: raises, promotions, development opportunities, job security. A bad review doesn't motivate anyone. It frustrates people or demoralizes them.

More importantly: specificity is what drives growth. "Be a better collaborator" doesn't change behavior. "In our last project, when the designer suggested a different approach, you pushed back immediately instead of asking questions first. Try asking 'what's the thinking behind this?' before disagreeing" changes behavior.

AI defaults to generic praise. Your job is to make reviews specific. This is the difference between HR that matters and HR that's compliance theater.

The stakes are high. A person's performance review affects their motivation, compensation, growth opportunities, and perception of whether the company values them. If you're going to take the time to write a review, make it specific and actionable.

The Review Process: Gathering Specifics First

Before you ask AI to write anything, you need specifics. This is the most important step, and it's where most reviews fail.

What you need for a good review:
- 3-5 specific accomplishments (with context, impact, and business result)
- 1-2 growth areas (with real examples of when this showed up, not general weakness)
- 1-2 development actions for the next period (specific skills to work on)
- How they're performing relative to expectations for their role and level
- Trajectory (are they ready for promotion? On track? Below expectations?)

Most managers don't have this. They have impressions. "They were solid." That's not enough. Impressions vary; specifics stick.

How to gather specifics: The Manager Notes Template

Email your managers 2-3 weeks before reviews are due with a structured template:

"Describe [person]'s performance this year. I need specifics for a meaningful review.

Accomplishments: What are 3-4 things they did really well this year? For each, include: what they did, why it mattered, what was the outcome. Examples: 'Shipped the customer dashboard feature on time despite design delays. This was important because it unblocked our big customer renewal. Result: customer stayed with us.' Or: 'Mentored two junior engineers, and both got promoted to mid-level roles. This freed up my time for architecture work.'

Growth areas: What 1-2 things could they improve? For each, include a specific example of when this was an issue. Examples: 'Sometimes doesn't ask for help early, waits until stuck. Example: spent a week on problem X before asking for input, then got unstuck in 30 mins of discussion.' Or: 'Needs to improve presentation skills. Example: Q3 product review presentation was technical but missed the business case and stakeholder was confused.'

Development priorities: Based on their performance and growth potential, what should they focus on in the next review period? Example: 'Ready for tech lead role; should focus on code review leadership and architectural thinking.'

Role-level assessment: How are they performing relative to what we expect for a [title/level]? Are they exceeding expectations, meeting them, below, or in the middle? How has this changed over the year?"

Managers will give you bulleted notes. That's your input for AI, and it's gold. Specific examples are what make reviews matter.

Real Example of Specific Manager Notes

Instead of: "Sarah is a strong performer who collaborates well."

You need:
- Accomplishment: "Led the API redesign project. This was important because our old API was causing customer integration issues and we were at risk of losing two big customers. She coordinated with three teams (backend, product, sales), created a migration plan, and executed without service interruptions. Both at-risk customers renewed."
- Growth area: "Sometimes over-commits and then stresses about deadlines. Example: agreed to lead API redesign AND mentor two interns AND take on the compliance audit. By month two she was overwhelmed. We talked about prioritization and she delegated mentoring to someone else. She needs to get better at saying 'this is my top 3; I can't take on X right now.'"
- Development priority: "Ready for senior engineer or tech lead. Should focus on: visibility and communication (her work impacts other teams but she doesn't always tell them until late), and delegation (she hoards work instead of leveling the team)."

This is what good manager notes look like. Specific, with examples, showing growth area without being mean.

The AI Review Draft: Organization, Not Substance

You have specific notes from the manager. Now you ask AI to organize them into a well-written, balanced review.

The right prompt:

"Organize these manager notes into a performance review.

Structure:
1. Opening: Overall assessment (are they exceeding, meeting, or below expectations for their level?)
2. Accomplishments: Summarize each accomplishment with its impact. (Use the specific examples from the notes.)
3. Growth areas: Describe what they could improve with specific examples of when this showed up. Make it constructive, not harsh.
4. Development plan: What will they focus on in the next period? How will we support them?
5. Trajectory: Where are they headed? (promotion track, lateral growth, continue in role, etc.)

Tone: Balanced, specific, constructive, professional. Use the notes as source truth. Don't soften or generalize specifics.

Manager notes: [paste notes]"

AI will produce something organized and well-written. You review it for accuracy and tone.

Common AI failures to watch for:

• 
  AI softens language: Manager's note says "missed deadlines repeatedly." AI says "has opportunities to improve time management." Don't let it soften. If the manager said missed deadlines, the review should say that (constructively: "missed deadlines on X and Y projects; we talked about what got in the way; next period focus on estimating realistically and flagging blockers early").
• 
  AI over-generalizes: Manager notes specific example. AI generalizes. Manager note: "Spent a week on problem X before asking, then unstuck in 30 mins of discussion." AI says: "could be more collaborative." No. The review should mirror the manager's specificity: "sometimes waits too long to ask for help; example from Q2 when you spent a week on X before reaching out, then got unstuck in 30 minutes of pair discussion. Next period: flag when you're stuck after 2-3 hours, not after a week."
• 
  AI adds praise not in the notes: You said solid contributor. AI says "exceptional talent," "strong leader," "critical to the team." Don't let it embellish. You provided your assessment; use that.

Reviewing the AI draft:

• Read it and ask: Does this accurately reflect the manager's notes?
  - Check each accomplishment: Is the impact clear? Does it show why it mattered?
  - Check growth areas: Are they specific to this person's examples? Or did AI generalize?
  - Check development plan: Is it clear what they should work on?
  - Check tone: Is it professional, constructive, specific? Not harsh, not softened?

If it fails any of these, revise. Don't publish a first draft.

The Balanced Review: Accomplishments and Growth

Good reviews have both accomplishment and growth area sections. This isn't "sandwich feedback" (criticism surrounded by praise). It's an honest, balanced assessment.

Why both matter:

If you only praise, they don't know where to improve. They think they're perfect. Nothing changes.

If you only critique, they get demoralized. They think they're doing everything wrong. Motivation crashes.

If you balance (here's what you did well, here's where you can grow, here's how we're supporting your development), they understand: we value you, we see your contributions, we also see where you can improve, and we're invested in your growth.

How to structure it:

Opening/Overall Assessment:
"You had a strong year. You exceeded expectations on shipping (delivered 6 projects on schedule when we expected 4). You're meeting expectations on collaboration but have room to grow in cross-team communication."

Accomplishments section:
"What you did really well this year:
- [Accomplishment 1 with impact]
- [Accomplishment 2 with impact]
- [Accomplishment 3 with impact]"

Growth area section:
"Where you can improve:
- [Growth area with specific example and how they can improve]
- [Growth area with specific example and how they can improve]"

Development plan:
"For the next review period, we want you to focus on:
1. [Skill/area with specific actions]
2. [Skill/area with specific actions]

Here's how we'll support you: [coaching, training, project assignment, mentoring, etc.]"

Trajectory:
"You're on a [promotion track / continuing to grow in this role / lateral move] path. Next steps: [specific actions that position them for that path]."

This tells a complete story: they're good at X, they're growing in Y, and here's how we're developing them. They know where they stand and where they're headed.

Important: A review should answer four questions: How am I doing? Where am I excelling? Where do I need to grow? What's next for me? If the review doesn't answer all four, rewrite it.

The Role-Level Alignment: Comparing to Standard

A person's performance should be compared to what's expected for their role and level. This is critical because it's the baseline for everything: compensation, promotions, development.

Manager note: "She's a strong individual contributor."

But what does that mean? Is she exceeding expectations for her level? Meeting them? Below?

A strong individual contributor junior engineer is expected to: learn quickly, ship features with guidance, ask good questions, meet deadlines with support.

A strong individual contributor senior engineer is expected to: ship complex features independently, mentor others, think about architecture and long-term impact, solve ambiguous problems.

"Strong individual contributor" means very different things at different levels.

Role expectations framework:

Create a simple role expectations document for each level in your organization. Example:

Junior Software Engineer (Year 1-2)
- Expected: Learn systems, ship small features with guidance, write testable code, ask questions
- Exceeding: Ships medium features independently, improves systems, mentors interns
- Below: Takes more than 3 months to ship first feature, needs significant code review feedback

Mid-Level Software Engineer (Year 2-5)
- Expected: Ship features independently, improve code quality, mentor one junior, participate in design discussions
- Exceeding: Leads design decisions, improves architecture, mentors multiple people, reduces technical debt
- Below: Still needs close oversight, slow to ship, doesn't mentor

Senior Software Engineer (Year 5+)
- Expected: Lead architecture decisions, mentor multiple engineers, improve system reliability, drive long-term tech vision
- Exceeding: Define team's technical strategy, grow engineers to senior level, ship projects that unblock the organization
- Below: Still focused on individual contributions, doesn't grow others, reactive (not proactive) on tech direction

How to use AI:

"This person is a [title/level]. I want the review to compare their performance to what we expect at this level. They excelled at [X] which is a [junior/mid/senior] level expectation. They need to work on [Y] which is important for their level. Draft the review showing how they compare to role expectations."

Example:
"Sarah is a mid-level engineer. She excelled at shipping features independently (that's the expectation) and also at mentoring an intern (exceeding). She needs to work on participating in design discussions with confidence (that's a mid-level expectation she's below on). Draft a review that shows where she's strong, where she's on track, and where she needs to improve for a mid-level engineer."

This makes the review meaningful. It tells them where they stand relative to their role. "You're exceeding expectations" is motivating. "You're below expectations for a mid-level engineer on X" gives them clear direction for improvement.

Tip: Use role expectations as your review framework. If you don't have role expectations documented, create them. It makes evaluation consistent and fair.

The Development Plan: Making Growth Actionable

A good development plan turns feedback into action. "Get better at X" is not a plan. "Work on presentation skills by: taking a presentation class, presenting at two team meetings, getting feedback from your manager" is a plan.

How to structure a development plan:

For each growth area, specify:
1. What they're developing: Clear skill or area
2. Why it matters: How it will help them
3. How they'll develop it: Specific actions (training, projects, mentoring, etc.)
4. How you'll support it: What the organization will provide
5. How you'll measure it: How will we know they improved?
6. Timeline: When will you revisit this?

Example:

Growth area: Cross-team communication

Plan:
- What: Learn to communicate technical decisions to non-technical stakeholders
- Why: You're on the path to tech lead, which requires influence across the organization
- How: (a) Take a presentation class, (b) present monthly project updates to product team, (c) get feedback from your manager each time, (d) read [book on communication]
- Support: We'll pay for the class and give you time. Your manager will give feedback each month.
- Measure: By end of next review period, product team reports better understanding of your decisions. You feel more confident in these presentations.
- Timeline: Revisit in 6 months (at mid-year review).

This is a plan. They know what to do. You're committed to supporting them.

How to use AI to draft a development plan:

"Create a development plan for someone working on [growth area]. Include: what they're developing, why it matters, specific actions (projects, training, mentoring, etc.), how we'll support them, how we'll measure progress, and timeline. Make it actionable, not vague."

AI will draft something. You personalize it.

Common Review Mistakes and How to Fix Them

Mistake 1: Too generic
"Jane is a valued team member who contributes well."

Fix: Add specific examples. "Jane shipped the customer portal feature on schedule. This was critical because it unblocked our biggest customer. She also improved our testing practices, which reduced bugs by 20%."

Mistake 2: Only praise or only critique
"John is an excellent engineer. He's learned a lot this year."

Fix: Add growth areas. "John is an excellent engineer. He's learned a lot on [specific things]. He should work on [growth area] next period."

Mistake 3: Vague growth areas
"Sarah should work on communication."

Fix: Add specific example. "Sarah should work on communication. Example: in the Q3 project, she made a key decision without looping in product until late. Next period: involve key stakeholders earlier in decisions."

Mistake 4: No connection to development
"Mike is a strong performer. We hope to see continued growth."

Fix: Add specific plan. "Mike is a strong performer on X. He should focus on Y next period. We'll support this by: [specific actions]."

Mistake 5: Inconsistent with manager assessment
Review says "exceeds expectations" but no accomplishments justify it.

Fix: Check that accomplishments match assessment. If they exceed expectations, there should be examples of work beyond what's expected.

Try This Now: Four Exercises

Exercise 1: Gather Specific Manager Notes

Interview a manager (or be one). Ask about one of their reports who had a good year.

Ask: "Give me 3 specific things they did really well. For each, tell me: what did they do, why did it matter, what was the outcome?"

Write down these specifics. Then ask: "Give me 2 things they could improve. For each, give me a specific example of when this was an issue."

Document these specific examples. This is your review input.

Exercise 2: Create a Role Expectations Document

Pick a role in your organization. Document what you expect from someone at each level in this role:
- Junior/Early (Year 0-2): What are baseline expectations?
- Mid-level (Year 2-5): What are growth expectations?
- Senior (Year 5+): What are leadership expectations?

For each level, specify 3-4 key expectations. Example: "Mids should mentor one junior. Seniors should mentor multiple people."

Exercise 3: Review Draft from Notes

Ask AI: "Draft a performance review using this input: [paste specific manager notes]. Structure: overall performance, accomplishments (with specific examples), growth areas (with examples), development plan for next period. Tone: constructive and specific, not generic."

Review what AI produces. Is it specific? Does it capture what the manager said? Would the person reading this understand what they did well and where they should improve?

Exercise 4: Revise for Specificity

Take a recent review you've written (or drafted with AI). Ask yourself: Are all accomplishments specific with context? Are all growth areas specific with examples? Is there a development plan with specific actions? If not, revise to add specificity.

Practical Application - "What to Do Monday Morning"

• 
  Create a manager notes template: Make it easy for managers to give you the specifics you need for a good review.
• 
  Gather specifics before writing: 3-5 accomplishments with examples, 1-2 growth areas with examples, development priorities.
• 
  Use AI to organize, not create: Your notes + AI structure = good review. Not AI generating from thin air.
• 
  Ensure specificity: If AI softens or generalizes, rewrite to be specific. Specific beats generic every time.
• 
  Compare to role expectations: Create clear role expectations so every manager knows what "good" looks like at each level.
• 
  Include development action: What will they work on next period? How will you support them?
• 
  Share draft with manager: They provided the input; they should review the review before you finalize.

Key Takeaways

• Specifics beat praise: Generic compliments don't help anyone grow.
  - Gather examples before writing: Managers' notes drive the review; you organize them.
  - Balance accomplishment and growth: Both matter. Neither alone is a good review.
  - Compare to role expectations: Where do they stand relative to their level?
  - Include development action: What's next? How will you support it?
  - Use AI to organize, not create: AI is the writer; you're the strategist.
  - Make it actionable: If they read the review, can they understand what to do better?

FAQ

Q: How long should a review be?
A: 300-500 words is typical. Long enough to be specific (3-5 accomplishments, 1-2 growth areas, a development plan), short enough people actually read and remember it.

Q: Should I share drafts with managers before finalizing?
A: Yes. You're synthesizing their input. They should review to make sure you captured their assessment accurately.

Q: What if I disagree with the manager's assessment?
A: Discuss before finalizing. You might have different perspectives based on cross-team work or different contexts. It's important to align.

Q: Can I use AI to draft reviews I'm writing as a manager?
A: Yes, same process. You provide specifics; AI organizes. The notes are yours; AI just helps you articulate them well.

Q: What if a manager says 'I don't have enough specifics'?
A: That's common. Ask them: "Give me one project this person shipped. What did they do? What was the outcome?" Start there. Get concrete.

Q: How do I handle a review where performance is mixed (some good, some bad)?
A: Show both. "You excelled at X. You struggled with Y. Here's what you should focus on next period." Mixed is realistic; show the full picture.

Q: Can I use the same review structure for everyone?
A: The structure yes (accomplishments, growth areas, plan). The content should be completely personalized based on their actual performance.

What's Next

Lesson 5.2 is about goals: how to work with managers and employees to set goals that drive performance, not just fill a template.