The Exit Interview Nobody Reads

Almost every company runs exit interviews. Almost none of them can tell you what changed as a result of the last ten. The transcript goes into a folder, the folder goes into a drive, and the drive is opened next when someone needs to defend a termination.

This is a strange waste. Exit interviews are the only moment in the employment relationship where the incentive to be polite drops sharply and the person still has full context. The data quality is unusually good. The processing is unusually bad.

Fixing it doesn't require a new tool. It requires three changes: better questions, a pattern-tracking method instead of a filing system, and one recurring meeting where the patterns get read out loud.

Why the current version fails

Four structural problems, all fixable.

• The questions invite summaries. "Why are you leaving?" produces one sanitized headline — usually compensation or growth — that compresses a year of specific friction into a category.
• The output is prose. Twelve pages of narrative per departure cannot be compared to the last twelve pages, so nobody tries.
• The reader is the wrong person. HR conducts and files it; the manager whose behavior is most often the subject never sees it, and the executive who could change the system never asks.
• Nothing closes the loop. Because no action follows, current employees learn that honest feedback goes nowhere — which degrades the quality of every other channel you have.

Three questions most exit interviews skip

Keep whatever you're asking. Add these three, and ask them in this order, late in the conversation.

1. "When did you first start thinking about leaving, and what happened around then?"

The stated reason for leaving is usually the final trigger. This question locates the origin, which is almost always something specific and much earlier: a reorganization, a promotion decision, a project that was cancelled without explanation, a moment where they raised something and nothing happened.

The distance between the origin and the resignation is itself useful information. If people typically start thinking about leaving nine months before they resign, you have a nine-month window you're not using.

2. "What did you stop bringing up, and why?"

This is the highest-yield question in the set. Everyone who leaves has abandoned at least one topic. The answer names both an unresolved problem and the mechanism by which your organization suppresses information — those are two findings, from one question.

Expect answers like "the staffing model, because I raised it twice and got a lecture about priorities." That's not a grievance; that's a process defect with a named cause.

3. "What would you warn a friend about before they took your job?"

The friend framing does something the direct question can't: it gives permission to be blunt without making the person feel they're settling a score. The answers are consistently the most specific and most actionable in the transcript.

One optional fourth: "What should we absolutely not change?" It's not politeness — it identifies the load-bearing parts of your culture before someone helpfully reorganizes them away.

The pattern-tracking method

Stop storing transcripts as the primary artifact. Start maintaining one table — a spreadsheet is entirely sufficient — with a single row per departure and these columns:

1. 1Date, team, manager, tenure, level, and whether the exit was voluntary.
2. 2Stated reason — their headline, in their words.
3. 3Origin event — the answer to question one, in one sentence.
4. 4Abandoned topic — the answer to question two.
5. 5Warning to a friend — the answer to question three, verbatim where possible.
6. 6Category tags — pick from a fixed, short list: manager, decision rights, workload, growth path, pay, product direction, peer quality, process friction.
7. 7Regretted? — yes or no. This is the weighting column, and it changes everything.

The fixed tag list matters more than it looks. Free-text categories can't be counted, and the whole purpose is counting. Keep the list under ten items and resist adding a new tag until the same phrasing has appeared three times.

Two rules for reading the table:

• Weight regretted departures far more heavily. Three regretted exits tagged "decision rights" is an emergency. Three non-regretted exits tagged the same way may just be normal churn.
• Read by segment before reading in aggregate. Aggregated exit data almost always says "pay and growth," which is useless. Sliced by manager, team, and tenure band, it usually points somewhere specific and uncomfortable.

The quarterly review that makes it real

Once a quarter, forty-five minutes, on the leadership team's actual agenda — not a report circulated by email. The structure that works:

1. 1Read the "warning to a friend" column out loud, verbatim, for every departure that quarter. This takes five minutes and does more than any summary slide.
2. 2Show tag counts for the quarter and the trailing year, with regretted exits marked.
3. 3Name the top one or two patterns. Two is the maximum; a list of eight patterns produces zero changes.
4. 4Assign an owner and a date for each. "We'll be more thoughtful about growth paths" is not an action. "Karen publishes promotion criteria for the delivery track by March 15" is.
5. 5Review last quarter's actions first, before looking at new data. If the previous two didn't happen, don't add a third — fix why.

One necessary tension: managers named repeatedly in exit data need to hear it, and the departing employee needs to be protected. The workable rule is to share patterns, never transcripts, and to require at least two independent instances before raising it with a manager — then raise it as a coaching conversation about a pattern, not an accusation about a person.

Close the loop publicly

The last step is the one that converts exit data into culture change: once a quarter, tell the people who still work there that something changed because of what people said on the way out. One or two lines is enough — "we heard consistently that scope decisions had no clear owner, so here's the new rule."

Do that twice and you'll notice something: people start telling you these things before they resign. At which point the exit interview stops being an autopsy and starts being a backstop.