Exit interviews that produce action, not a file

A company I looked at had held forty-one exit interviews over two years. Every one was completed, signed and filed. When I asked what they had learnt from them, the HR executive said people mostly leave for better opportunities. That was true of the forms. It was not true of the company, where eleven of the forty-one had left one department of thirty people.

The interviews were not the problem. The problem was that nobody had ever put forty-one of them next to each other.

Three things that make an exit interview useless

• The leaver's own manager conducts it. Nobody tells their manager that their manager is the reason. This single choice removes the most valuable finding an exit interview can produce.
• It is held on the last day. The person has mentally left, has a settlement to protect, and wants to go. Anything they say will be polite.
• The findings are filed individually. One interview is an anecdote. Twenty are a finding. A process that never aggregates cannot produce the second.

Fix those three and the same conversation starts producing something usable, without any change to the questions.

Who, when and how

1. 1

   Conducted by HR or a neutral senior person

   Never the reporting manager. In a small business where HR reports to the person in question, use someone else senior and say plainly who will and will not see the notes.

2. 2

   Held mid-notice, not on the last day

   Far enough in that the person is still engaged and their memory is fresh, late enough that they are candid because the decision is already irreversible.

3. 3

   Confidential in a way you can actually honour

   Say exactly what happens to the notes — that findings are reported as themes, not attributed quotes. Then do that. One breach ends the usefulness of the process for everyone who hears about it.

4. 4

   Twenty to thirty minutes, structured

   Long enough to get past the first answer, short enough that it is not an ordeal. A structured format also makes aggregation possible, which an open conversation does not.

5. 5

   Notes written the same day

   Against a fixed set of fields so the twentieth interview can be compared with the first.

Ask about moments, not feelings

"Were you satisfied with your role?" produces a rating nobody can act on. "When did you first think about leaving, and what had just happened?" produces a date and an event. People remember events accurately and rate feelings vaguely, so the questions should be about events.

1. When did you first think about leaving? What had just happened?
2. What made you start looking, as opposed to just thinking about it?
3. What would have had to change for you to have stayed?
4. Which part of the job was different from what you expected
   when you joined?
5. How would you describe your manager's strengths? And where
   did you most often need something you did not get?
6. What did we do well that we should not change?
7. Would you recommend working here to someone you rate?
   Why, or why not?
8. Is there anything you would say now that you would not have
   said three months ago?

Record: department, manager, tenure, role, grade, joining date,
last working day, regretted / not regretted.

Question one does most of the work. The gap between the triggering event and the resignation is usually three to six months, and that gap is where a business could have intervened if anybody had known.

Question eight is the one that surprises people. Asked plainly, near the end, it frequently produces the thing the person came in intending not to say.

Aggregate, or do not bother

Every quarter, put the interviews side by side and cut them the ways that reveal something: by manager, by tenure band, by department, by role. You are looking for concentration, not for the most common answer.

• Concentration by manager. Four exits from one manager in a year, in a company where the average is one, is a finding no individual interview would have produced.
• Concentration by tenure. People leaving at four to six months is an onboarding or a hiring problem. People leaving at two to three years is a progression problem. The remedies share nothing.
• Recurring triggering events. If the same event type appears in eight interviews, it is a process, not a coincidence.
• The gap between stated and real reasons. Compare what the form says with what the notes say. "Better opportunity" on the form and a manager story in the notes is the most common divergence there is.

What to do when the finding is a manager

This is the finding businesses most often have and least often act on, because acting on it is uncomfortable and the evidence feels soft. Two things make it manageable. First, present the concentration rather than the quotes — four exits from one team against an average of one is a number, not an accusation. Second, treat it as a capability question before a performance one. Most managers producing this pattern were promoted for being good at the work and never shown how to run people.

That is a training question, and it is the specific gap leadership and people management work addresses — delegation, feedback and the conversations a first-time manager avoids until they become resignations.

The interview you are not holding

The honest limitation of exit interviews is that by the time you hold one, the decision is made and usually cannot be reversed. Everything you learn is applied to the next person, not to this one.

The conversation with more leverage is the one held with someone who is not leaving: what would make you consider it, what is frustrating, what would have to change. Businesses rarely run these because they feel like inviting the problem. In practice they surface exactly the issues the exit interviews will surface in nine months' time, while there is still something to be done about them.

Both belong in the same piece of work as the rest of the exit and offboarding process — the sequence, the clearance, the settlement and the feedback — because they share the same weakness: each step is done reasonably and nothing joins them together.

Close the loop, visibly

One last thing, and it is what separates a process that keeps working from one that quietly dies. When a finding leads to a change, say so internally — not naming anyone, but stating that a change was made because of what leavers told us. People participate honestly in a process they have seen produce something, and stop participating in one they have not.

And where the same exit reasons keep appearing across quarters, the question has moved beyond exits. That is when it is worth measuring properly — who leaves, when, from where — which is what employee retention and attrition management work is for.

One measure worth tracking

Count how many leavers accept the interview and how many decline. The rate itself is unremarkable; the distribution is not. A run of declines from one department, or from people reporting to one manager, is a finding before a single question has been asked.

Track it quarterly alongside the themes. It costs nothing, and it is the one number in this whole process that cannot be influenced by how well the interviewer asks questions.

Where the declines and the themes point the same way, you have as close to evidence as this kind of work produces — and that is usually the moment a business is willing to act on something it had previously treated as a matter of opinion. Handled as part of the wider exit and offboarding process, the interview stops being an administrative step at the end and becomes the only structured feedback a departing employee ever gives you.