Why People Actually Left
Use digital records to locate a process problem, not to make an unsupported judgment about a person. That distinction should be written into any evaluation of activity reporting software and reflected in who can access the resulting data.
When this issue touches live workplace policy, compare the local evidence with guidance published by CIPD before changing a process or communicating a rule.
The only place people tell you the truth, collected badly almost everywhere, and the categories that make the data useless.
Every operation records leaving reasons. Almost none of them records anything usable.
Why the data is bad
The categories are written for payroll: resigned, dismissed, end of contract.
Which tells you the mechanism and nothing about the cause.
The conversation, where it happens, is with the person's supervisor — who may be the reason they are leaving.
It happens on the last day, when the person wants to go and has no incentive to be candid.
And "personal reasons" absorbs everything, which is the single largest category in most systems and means nothing.
When to ask
Not on the last day.
A week or two after they have gone, by phone or a short form, when there is nothing to lose and the new job has started.
Response rates are lower and the answers are far better, which is the right trade.
And ask the first-fortnight leavers especially, who are the largest group and the least often asked because they barely registered.
Who should ask
Not their supervisor.
Somebody neutral, who can say plainly that the answers will not affect a reference.
And who will actually pass on what they hear, including the parts that are about a named manager.
Categories that produce usable data
Pay or hours at another employer.
Shift pattern did not fit.
Transport.
Physical demand.
Relationship with a manager or colleague.
The job was not as described.
Progression or lack of it.
Health, including work-related.
Personal circumstances genuinely unrelated to work.
Nine categories, and the last one should be a small minority — if it is the largest, the categories are not being used.
Reading it honestly
Look for concentration: one shift, one team, one task, one time window.
Compare first-fortnight leavers against six-month leavers, because the causes are entirely different and averaging them produces a picture of neither.
And compare what leavers say against what current staff say, which is a separate exercise and frequently contradicts it.
The finding nobody wants to write down
A large share of exit data points at supervisors and at the accuracy of the advert.
Both are uncomfortable to report upward, and both are cheap to fix relative to a pay increase.
A report that omits them because they are awkward has wasted the whole exercise, and everybody who was asked will notice that nothing changed.
Acting on it
One finding per quarter, acted on visibly.
Told to the current workforce: people left because of this, we changed that.
Which is what makes the next round of answers honest, and the absence of it is why exit data everywhere is bland.
The measure
Response rate to post-exit contact.
Proportion in the "personal reasons" bucket, which should fall as the process improves.
Findings acted on against findings identified.
And attrition by the cause category over time, which is the only way to know whether anything you did worked.
Asking the people who stayed
Exit data tells you why people left. It cannot tell you why the others stayed.
Ask a sample of six-month and two-year staff what nearly made them leave and what stopped them.
The answers are consistently specific — a supervisor who was reasonable about a childcare problem, a shift swap that was approved, a transfer to a different task.
Which are all things an operation can do deliberately rather than by luck, and none of them appears in any exit report.
Keeping the door open
A good leaver is a returner and a referrer.
Which means the last two weeks matter: notice worked properly, a reference given, a thank you.
And an explicit statement that they would be welcome back, which costs nothing and produces a measurable share of rehires in operations that do it.
Most warehouses treat resignation as a betrayal for the remaining fortnight, and then wonder why nobody returns.
How it fails in practice
The conversation is held on the last day by the person being left, and the answer is recorded as personal reasons.
The data then shows that nothing the operation controls is causing anybody to leave.
The check
Take twenty leavers from last quarter and call them.
The difference between what they say and what the system recorded is the measure of how much the current process is worth.