Exit interviews: questions, recording review and an evidence-based action plan
Conduct an exit interview and review the recording: voluntary participation, questions about specific events, facts and opinions, anonymization and a plan for changes. Includes a guide.
The conversation helps you understand the employee’s experience
A person rarely leaves for a reason that fits in one sentence. Tasks, relationships, expectations and circumstances outside the company may all matter. An exit interview can identify working conditions the company can improve, without assigning blame or trying to confirm a preferred explanation.
Tell the participant who can see the recording and findings, whether they have to answer and whether they can skip a question. Data protection and retention requirements vary. Review them with the responsible specialists. Avoid promising anonymity when details make the role easy to identify.
Set safe ground rules and make conclusions traceable
Agree on the interview rules
Before recording, explain the purpose, participants, recording method, access and retention under approved policy. Obtain any necessary consent. Let the person decline recording or skip a question. The organization checks its procedure with the appropriate specialists.
Ask about events and expectations
Start with an open account. Ask what happened, what the person expected, whom they contacted and what changed. Avoid questions that contain an accusation. Collect third-party details only if necessary to understand the process.
Analyze the material in layers
Mark events, personal assessments, needs and suggestions in the transcript. Keep exact words where nuance matters and preserve the context of any quote. One account gives you a lead to check, not statistical evidence of a general problem.
Assign follow-up to process owners
State the observation neutrally, with its basis and confidence level. Name who will check it, when and how feedback will be handled safely. Remove combinations of details that unnecessarily identify a person in an aggregate review.
Review of a fictional comment about feedback
This is a fictional training example. The names, circumstances and outcomes are included solely to explain the method.
I brought in a draft three times. All I heard was “redo it.” In the end I couldn’t tell what counted as good work.
This suggests a testable concern: the quality criteria may have been unclear. One account cannot establish that the manager always ignored the employee or that this was the only reason for leaving.
Check what criteria exist for the task and how they are used in other cases. A neutral paraphrase without a name or unnecessary details is enough for the report. The scenario and quote are fictional.
| Layer | Example entry | Status |
|---|---|---|
| Event | Three drafts were returned | Reported by participant |
| Assessment | The criterion was unclear | Interpretation of the experience |
| Need | Specific feedback | Hypothesis to check |
| Action | Review the task’s criteria | Owner’s decision |
Exit interview guide and review worksheet
Fill in only the fields needed for the decision. Store access restrictions and deletion dates separately.
Purpose of the conversation: Recording and use rules: Consent / restrictions: What was expected of the role: Specific difficult episode: What the employee did: Response they received: What might have helped: Fact / assessment / need: What needs checking: Safe aggregate finding: Owner of the next action:
Check accuracy and ethical handling before sharing findings
- Voluntary participation and data use were explained.
- Quotes include context that affects their meaning.
- A single account is marked as a signal to check.
- The report contains no unnecessary identifying details.
A closer look
Should we send the questions in advance?
That can help the participant recall facts and feel more comfortable. Leave room for an open conversation and don’t require written answers.
Can we compare reasons for leaving by category?
Yes, when categories are consistent and there’s enough data. Keep the original context and avoid presenting a small sample as a picture of the whole company.