People & HR

Exit interview

Get the real reason for leaving by asking once, plainly, after the decision is already final.

Why this survey gets sent

“Everyone writes ‘personal reasons’ on the way out, and six months later I am recruiting for the same seat for the third time.”

People leave for a reason they will say exactly once, and only to someone who no longer stands to lose anything by hearing it. A written survey — sent after the notice is accepted, and read by the people team rather than by the manager being described — gets an answer that a meeting with that same manager never will.

What is in it

  • The main reason for leaving (single response)
  • If it was another reason, what was it? (short answer)
  • What would have changed your mind? (long answer)
  • Manager, the work, pay, growth, decisions — one rating grid
  • Would you recommend us? Would you come back?
  • What the next person needs to know (long answer)
  • Length of service and last working day

About 5 min to answer

Exit interview

Use this template

How to write this survey

Send it after the notice, not during it

Ask while the resignation is still being discussed and what you get is a negotiating position, not a reason. Ask after the last working day and you get nothing at all: the account is closed, the address bounces, and the person has moved on. The week after the notice is accepted is the window — the decision is final, the memory is intact, and there is nothing left to lose by being exact.

A second, much shorter survey ninety days later is worth building into the process. Somebody who has started somewhere else is answering a different question, because they finally have a comparison, and the sentence “my new manager does this every Monday morning” is worth more than any rating they gave you on the way out.

Give the real reason somewhere to go

“Personal reasons” is what someone writes when the honest option is not on the list. A list of reasons earns its place only if it contains the answers that embarrass you: my manager, the pay, there was nowhere to go next, the job was not the one I was hired into. Leave those out and you have built an instrument that can only confirm what you already believed.

Ask for one main reason rather than “tick all that apply”. People leave for four reasons and act on one, and the single choice is what tells you which one crossed the line. Put a short text box directly beneath it for the answers your list could not predict, and read the two together — the box is usually where the sentence you will quote at the leadership meeting is sitting.

Rate the conditions, not the person

A grid covering the manager, the work itself, the pay, the chances to grow and the way decisions were explained gives you four or five comparable numbers per leaver — the only part of an exit survey you can add up. Word every row as a condition: “the support you had from your manager”, not a verdict on a colleague. The first can be repaired; the second is a complaint file, and it will be read as one.

Two closing questions carry more weight than the grid above them: would you recommend us to someone looking for work, and would you come back one day? A leaver who says yes to both is the cheapest hire you will ever make and belongs on a list somebody actually keeps. A leaver who says no to both while rating every row well is telling you the problem is not on your list at all — and the open box is where they will say what it is.

Say who reads it, and mean it

An exit survey is not anonymous — you know exactly who is leaving — so anonymity is not the promise to make. The promise is about circulation: this goes to the people team, it does not reach your line manager in your words, and it has no effect on your reference or your final settlement. Write those three sentences into the intro and keep them literally.

Keeping them means aggregating before anything is shared. A manager is entitled to know that three of the last five leavers rated growth as poor; nobody is entitled to a sentence in someone's own handwriting. In a team of two, even a count identifies its author — hold those results until the next quarter and report them with everybody else's, or do not report them.

The handover box, and reading six of these at once

One question pays for the whole thing on its own: what should the person taking over know in their first week, including anything unfinished or waiting on a decision? It is asked at the only moment when somebody has both the knowledge and a reason to write it down, and it is the difference between a replacement who is useful in month one and a replacement who is useful in month four.

Then read exits in batches and never one at a time. A single resignation is a story about a person, and a story invites another story back. Six of them read together in one sitting are a pattern about a team, a pay band or a manager — and a pattern is something a leadership meeting can act on without anyone needing to be blamed for it.

Make this template your own survey.