When performance documentation is weak, it usually has one of two problems. It either says too little - "not meeting expectations" - or it says too much in language that is subjective, emotional, or difficult to support.
Experienced HR professionals know the difference between a useful record and a risky one. The goal is not to make every note sound legalistic. The goal is to record what happened, why it mattered, what was communicated, and what needs to happen next.
The problem with vague documentation
Consider this common sentence: "Jordan has a poor attitude and is not reliable." It may reflect the manager's frustration, but it does not tell the reader what happened. Was the issue missed deadlines? Attendance? Refusal to follow instructions? Tone in meetings? Something else?
Employee has a bad attitude and is not committed.
On three occasions in June, the employee missed the agreed 10 a.m. status update without notifying the project lead. The delay required the lead to follow up directly and slowed the weekly client summary.
The improved version is not perfect, and it may still need policy or HR review. But it separates observable behavior from interpretation.
A fact-first documentation framework
Before drafting, gather the following:
- Expectation: What policy, role requirement, goal, or instruction applied?
- Observation: What happened, when, and how do we know?
- Impact: What was the effect on work, team, customer, safety, compliance, or operations?
- Prior communication: What coaching, clarification, or warning already occurred?
- Employee response: What did the employee say or provide, if anything?
- Next expectation: What needs to change, by when, and how will it be evaluated?
Where AI can help
AI is useful for turning rough notes into a clearer first draft, spotting subjective wording, and organizing facts into a consistent structure. It should not decide whether discipline is warranted, determine credibility, draw legal conclusions, or invent missing details.
Copy the prompt below. Replace only the teal field with anonymized facts, then paste it into your organization's approved AI tool.
1. Goal
Help me turn the anonymized notes below into a fact-based documentation draft for HR review.
2. Add your information
- Anonymized notes: [paste anonymized facts]
3. Writing direction
- Separate observable facts from interpretations.
- Use neutral workplace language.
- Do not invent missing facts.
- Flag assumptions, policy questions, and any areas that may require HR or legal guidance.
4. Improve the result
After I submit this initial prompt—and before drafting—ask me three focused questions that will help you produce a more specific, accurate, and useful output.
A common situation
A manager reports that an employee is "not responsive." Before drafting anything, ask what that means. Did the employee miss internal response-time expectations? Fail to respond to a customer? Ignore a direct instruction? The documentation should reflect the specific workplace expectation, not the manager's label.
Human review checklist
- Does the document use observable behavior instead of labels?
- Are dates, expectations, and impacts accurate?
- Does it avoid diagnosing motives, personality, or intent?
- Does it include the employee's response where appropriate?
- Does it align with policy, past practice, and the level of concern?
- Does anything require HR, privacy, security, or legal review before use?
FAQ
Can AI make documentation risk-free?
No. Avoid that claim. AI can help structure a fact-based first draft, but defensibility depends on facts, policy, consistency, process, and appropriate review.
Should I include every detail?
No. Include relevant facts and context. Avoid unnecessary personal, medical, protected, or speculative information.
What is the fastest improvement managers can make?
Replace labels with behavior. Ask, "What did the person do or not do, and what was the workplace impact?"