How to calculate employee turnover (and why you have three different figures)

Echipa HR 365 · reviewed 2026-09-02 · 6 min read

Employee turnover is leavers divided by the average headcount for the period, times 100. The part most often got wrong is not the formula but the denominator: average headcount, not the year-end figure. In a growing company, the difference can hide a third of the problem.

The formula, and why the denominator matters more than the numerator

Average headcount is obtained by adding the number of employees at the start of the period to the number at the end and dividing by two. For a stable company it barely matters which you use. For one that grew from 100 to 160 people, the choice changes the result by over 20%: the same 18 leavers give 11.3% against 160 and 13.8% against the average headcount of 130. The first figure is the one that usually reaches the leadership presentation, because it is smaller and easier to pull from a system.

If headcount varied a lot during the year — seasonal staff, a wave of hires in the third quarter — the arithmetic mean of the two endpoints no longer describes reality. Then you use the monthly average: add the headcount at the end of each month and divide by twelve. It is more work once and it removes a whole category of arguments about the figure.

Three formulas, three different questions

What we call “turnover” in everyday conversation is actually three indicators answering different things. Reported separately, they show you not just how big the problem is, but where it is.

The 16.4% figure alone says nothing. The three together say that nearly one in five people hired does not reach a year, while the rest of the organisation is relatively stable. That moves the conversation from “how do we keep people” to “who are we hiring and what happens to them in the first months” — two different budgets, two different owners.

The cuts that change the conclusion

A company-wide figure is useful only as an alarm. To become action, it needs three cuts, in this order:

1. By department or location — the company figure is almost always the average of two or three very different zones. One shop at 40% and the rest of the network at 12% together give 16%, a figure that describes nobody.
2. By tenure — under 3 months, 3–12 months, over a year. That distribution alone tells you whether the problem is recruitment, onboarding or the work itself.
3. By direct manager — the most uncomfortable cut and the most often avoided. Done on teams of at least 8 people, or a single departure produces absurd percentages.

A fourth cut, by reason for leaving, is useful only if the reasons are collected seriously. A field with predefined options filled in by HR at exit produces “personal reasons” in most cases, which is zero information.

How much is a lot?

There is no universal threshold, and anybody who gives you one is selling something. Normal turnover in a call centre or in retail is several times higher than in an engineering firm, and that says nothing about the quality of management in either. The only comparisons that produce decisions are with yourself: the same company, the same departments, the same period last year.

Comparing with the equivalent period is mandatory, not optional. Turnover has strong seasonality — January and September look different from July in almost any industry. A quarter compared with the previous one, rather than with the same quarter last year, produces false conclusions in both directions.

What you do once you have the figure

The monthly routine that turns the indicator into a decision

1. HR — calculates the three figures and cuts them by department and tenure, on the same day of the month, every time
2. HR — compares with the same month last year, not with last month
3. Department manager — receives only their own team’s figure and the tenure distribution, not the whole report
4. HR and manager — discuss one question: what changed before the departures, not why each person left
5. HR — notes the intervention and the date, so that in six months there is something to compare against

The last step is the one missing in most companies. Without a written trace of the intervention, any later improvement gets attributed to whoever speaks most convincingly in the meeting, and any deterioration stays unexplained.

Does the probation period count in turnover?

Yes, if the person was employed. Excluding probation departures makes the figure look better and hides exactly the problem the first-year indicator is looking for. You can report them separately, but do not take them out of the total.

What about seasonal staff, who leave at the end of the season anyway?

They are reported on a separate line, with their own average headcount. Mixed in with permanent staff, they double the company figure and make it useless for any decision.

Should I report monthly or annually?

Calculate monthly to see the trend, report quarterly to leadership. Over one month, in a small team, a single departure produces a percentage that frightens people for no reason.