YOUR TURNOVER RATE IS HIDING THE THING YOU NEED TO FIX

A single organisation-wide turnover figure averages together departures that have nothing in common. Four splits turn a number you report into a problem you can act on.

Emmanuel Duncan · 30 August 2026 · 6 min read

Most HR reports carry one turnover figure. It sits near the top of the people slide, it moves a little each quarter, and when a board member asks what is behind the movement, the honest answer is usually that nobody knows.

That is not a failure of effort. It is a property of the number. Turnover is a ratio built by adding together events that have almost nothing in common: the graduate who left after seven months, the supervisor who was dismissed, the senior accountant who retired, the storekeeper who followed a spouse to another region. Averaging those four together produces a figure that describes none of them.

The fix is not a better formula. It is four splits, each of which can be produced from records you already hold.

Split One: Departures You Regret

Classify every exit as regretted or not regretted. A regretted exit is one where you would have kept the person if you could. A non-regretted exit covers dismissals, exits at the end of a performance process, and departures where the role was ending anyway.

This split is the first one to build because it changes the meaning of the headline. An organisation losing twelve percent of its staff, where nine of those points come from a deliberate performance clean-up, is in a different position from one losing eight percent where every departure was someone it wanted to keep.

Two cautions. First, the classification has to be made by the line manager, not by HR, and it has to be made at the time of exit rather than reconstructed a year later. Second, it needs a rule, otherwise every exit becomes regretted by default. The rule I use is simple: would you rehire this person into the same role tomorrow.

Split Two: Tenure at Exit

Group departures into bands. Under six months, six to twelve months, one to three years, three years and over.

Each band points at a different owner and a different fix.

Departures under six months are a hiring and induction problem. Something in the recruitment conversation described a job that the person did not find when they arrived, or the first weeks left them without the tools, the clarity, or the supervision to get started. The people who can fix that are the hiring manager and whoever runs induction.

Departures between six and twelve months usually point at the manager relationship or at pay that turned out to be below market once the person had a comparison. Departures after three years more often reflect the absence of a next step.

If your exits cluster in the first band, no amount of work on retention benefits or long service recognition will touch the problem, because the people leaving never got far enough to care about either.

Split Three: Unit and Manager

Turnover is rarely spread evenly. It concentrates. In most organisations I have looked at, a small number of teams account for a disproportionate share of regretted exits, and the pattern holds across years rather than being a bad run.

Break the figure down by department, then by manager where team sizes allow. Then hold the result loosely, because a high figure can have three quite different causes: a manager who is genuinely difficult to work for, a team doing work that is unpleasant or physically demanding, or a team that has been used as the entry point into the organisation and is therefore always losing people upward.

The number tells you where to look. The conversation tells you what you found.

Split Four: The Quality of Your Exit Reasons

Pull the last fifty exit records and read the stated reason for each. Count how many fall into categories that carry no information: "personal reasons", "better opportunity", "resigned", "left".

If most of them do, you have an exit process producing paperwork rather than intelligence, and the first three splits are the only diagnostic you have. That is worth knowing before you build anything on top of the reason field.

Improving it is less about the form and more about who holds the conversation and when. Exit conversations run by the departing person's own manager collect politeness. Conversations run by someone else, a fortnight after the last day rather than during the notice period, collect more.

Working With Small Numbers

Many of the organisations I work with in Ghana have between forty and four hundred staff. At that size the percentages jump around, and a single quarter where three people leave a team of nine produces a headline figure that means nothing.

Three adjustments help.

Report counts alongside rates. Nine departures out of eighty is more legible than 11.25 percent, and it stops people over-reading a decimal point.

Use rolling twelve months rather than quarters. The rolling window absorbs the lumpiness while still moving when something real changes.

Set a floor for reporting by unit. If a team has fewer than eight people, report its exits as counts only and never as a rate, and never publish a manager-level rate that would identify an individual.

What This Changes in the Report

The single line becomes a short table. Total exits, split into regretted and not regretted, then regretted exits broken down by tenure band, with the two or three units carrying the concentration named underneath.

That table takes about the same space as the line it replaces, and it supports a decision. A board looking at it can ask whether the induction problem or the manager problem is the one being funded this year. A board looking at "attrition: 11.2 percent, up from 9.8" can only ask why it went up, which is the question nobody in the room can answer.

Where to Start This Quarter

Take your last twelve months of exits. You do not need a system change or a new tool for any of this, and in most cases the whole exercise fits in a spreadsheet.

Add three columns: regretted or not, tenure band, unit. Fill them in with the line managers, one department at a time, and accept that the classifications for exits more than six months old will be rough.

Then look at where the regretted exits concentrate, in tenure and in unit. That intersection, the band and the team where you are losing people you wanted to keep, is the thing your single turnover figure has been hiding.

Fix that, and the headline number will move on its own.

Emmanuel Duncan is the founder of Aletheia Lunara Consult, a professional education and consulting practice working with organisations across Ghana and beyond on people analytics, workforce research and organisational capability.