Strategies and practices used to keep valuable employees within an organization.
Employee retention is the share of employees who remain in post over a defined period: the people from the starting headcount who are still employed at the end, divided by that starting headcount. It is reported as a percentage, and it is not the arithmetic opposite of turnover, because it excludes anyone hired during the period.
Retention is always measured against a fixed starting cohort. Choose a period — a rolling twelve months is the most common — list everyone employed on day one, then count how many of those same people are still employed on the last day. Anyone hired during the period is not in the denominator and never enters the calculation, however long they stay.
Worked example: an organisation starts the year with 200 employees. Thirty of them resign or are dismissed before it ends, so 170 of the original 200 are still in post. Retention is 170 ÷ 200 = 0.85, or 85 per cent. The rate can never exceed 100 per cent, because the numerator is a subset of the denominator.
Retention rate (%) = (Employees from the starting cohort still employed at the end of the period ÷ Employees at the start of the period) × 100
Continue the example. The same organisation hired 40 people during the year, and 10 of those joiners left again before it ended. Total exits were 40 — the 30 originals plus the 10 new hires. Headcount finished where it began, at 200, so turnover measured the conventional way, against average headcount, is 40 ÷ 200 = 20 per cent. Retention was 85 per cent. The two figures do not sum to 100, and nothing has been calculated wrongly.
The gap is structural. The two metrics use different denominators and count different objects. Turnover counts events — every exit — against average headcount, so one unstable post filled and vacated three times contributes three exits, and turnover can exceed 100 per cent. Retention counts people, once each, against a fixed opening population, and is capped at 100 per cent. Reporting '100 minus turnover' as retention is therefore wrong whenever there is meaningful hiring, and the direction of the error depends on where the churn sat: exits concentrated among people hired inside the period inflate turnover while leaving the retention rate untouched.
85 per cent is not a number anyone can act on. Split the same 30 departures and the organisation looks different: if 12 came from one 20-person team, that team retained 8 of 20, or 40 per cent, while the remaining 180 people retained 162, or 90 per cent. The headline is an average of a crisis and a healthy business, and it describes neither.
First-year attrition is invisible in the headline by construction. The 10 joiners who left sit outside the starting cohort, so they cannot move the retention rate at all — yet losing 10 of 40 hires is a 75 per cent first-year retention rate, and that usually points at selection, onboarding or a job advert that oversold the role. Segment by tenure band, team, manager and role criticality, and separate resignations from dismissals and redundancies: an unsegmented rate treats the departure of the only person who can run the month-end close and a dismissal for misconduct as identical events.
Hold the period, the population and the treatment of leavers constant, or year-on-year comparison means nothing. Short windows flatter: over a quarter, people have had a quarter of the chance to leave, so a quarterly figure must never be read as an annual one. Decide once whether fixed-term, seasonal and probationary staff are in scope. Report involuntary exits separately, because a restructure can move the rate several points while saying nothing about whether people want to stay. And pair the rate with a length-of-service measure — an organisation can retain 85 per cent of its people and still lose most of its institutional memory, if the leavers are the long-tenured ones.
No. Retention measures a fixed cohort — the people employed on the first day of the period — against itself, so it cannot exceed 100 per cent. Turnover counts every exit during the period against average headcount, includes people who were both hired and lost inside that period, and can exceed 100 per cent when a single post churns repeatedly. Subtracting turnover from 100 gives the wrong retention figure in any period with meaningful hiring, and the size of the error grows with the hiring rate.
It means nobody employed at the start of the period left, which is worth understanding rather than automatically celebrating. It says nothing about people hired and lost during the period, because they were never in the cohort being measured. Sustained perfect retention in one team can also signal that nobody is moving — no promotions out, no fresh capability in — so read it alongside first-year attrition and internal mobility rather than on its own.
At minimum by tenure band, by team or manager, and by exit type. Tenure exposes first-year attrition, which the headline rate cannot show at all. Team-level rates reveal whether a mediocre organisation-wide number is actually one struggling group surrounded by healthy ones. Separating voluntary resignations from dismissals and redundancies stops involuntary exits either masking or manufacturing a retention problem. Where roles are hard to refill, track those roles separately instead of trusting the average.
Explore our full platform with a 14-day free trial on Standard. Manage employees, post jobs, hire faster, and manage your tasks effortlessly with our all-in-one platform.