It is 9 PM on a Sunday and you have six performance reviews due Monday morning. You know these people. You have opinions. But when you open the blank doc, nothing comes out. So you write the same vague paragraphs you wrote last year and feel slightly guilty about it.
Most managers write reviews by memory, which means they write about the last two weeks and call it a year. It is not dishonest — it is just how human memory works.
There is a different approach. It takes 12 minutes per review, produces something specific enough to actually help your employee, and you can do all six before the game ends tonight.
Here is the exact workflow.
What You Need Before You Start
- Any notes you have on the person: 1:1 notes, Slack messages, project outcomes, even a messy bullet list
- Access to any LLM: Claude, ChatGPT, or Gemini all work (Claude is best for this)
- 12 minutes per review
Paste this prompt into your LLM of choice. Replace the brackets with your specifics.
I am a [job title] writing a performance review for [employee name], a [their job title] who has worked for me for [X months/years]. Here is what I remember about their work this year: [PASTE: any notes, emails, Slack messages, 1:1 notes, project outcomes — even a messy bullet list works] Based on this, identify: (1) their 3 strongest contributions, (2) 2 areas where they fell short of expectations, (3) 1 skill gap that is limiting their growth. Be specific. Cite examples from what I gave you.
What happens: The AI synthesizes your scattered notes into a structured analysis. It will find patterns you forgot about — the project from March, the mentoring moment in June, the deadline miss in August.
The messier your input, the better this works. Do not clean up your notes. Just paste everything. The AI will organize it.
Now write a performance review for [name] using the analysis above. Format: 3 paragraphs. Paragraph 1: overall performance summary. Paragraph 2: specific strengths with examples. Paragraph 3: growth areas framed as opportunity, not criticism. Tone: direct, fair, professional. Avoid corporate HR jargon. Do not use the words "synergy," "bandwidth," or "leverage" as a verb. Length: 300–400 words.
What happens: You get a draft that sounds like you on a good day — specific, professional, and fair. Read through it once and make edits. Most people change 2–3 sentences.
I am considering rating [name] as [Exceeds/Meets/Below] expectations. Based on what I have shared, make the case FOR this rating and the case AGAINST it. Then give me your recommendation.
What happens: The AI gives you a balanced perspective on your rating. This is the step most people skip — and it is the one that prevents you from under-rating your quiet performers or over-rating your loudest ones.
Read the full draft once. Ask yourself:
- Would I say this to their face? (If not, rewrite that sentence.)
- Is there anything here their skip-level would find surprising? (If so, add context.)
- Did I include at least one thing they will be proud of? (Not optional.)
Done.
The Output
A 300–400 word performance review that is specific, balanced, and defensible. Your employee will read it and think you spent an hour on it. Your HR department will think you finally took the training seriously.
Your Action This Week
Open your calendar. Find the next performance review cycle — whether it is next week or next quarter. Block 90 minutes. That is enough time for 6 reviews using this workflow.
If your company does not do formal reviews, use this workflow for your next 1:1 prep. The same prompts work — just change "performance review" to "1:1 preparation notes."