Guides
How to calculate staff turnover: a UK employer's guide
Ask two people in the same business what the staff turnover rate is and you will often get two different numbers. HR counts the people who resigned. Payroll counts the leaving dates that reached HMRC. Both are working carefully, and both are right, because nobody agreed what a leaver was or what to divide it by.
That matters more for a shift workforce than almost anywhere else. When a third of your people are casual, bank or agency, and headcount doubles for the summer, the choice of denominator can move the published rate by ten points without a single extra person walking out.
The quick answer
Staff turnover is the number of leavers in a period divided by the average number of people employed over that same period, multiplied by 100. The formula is not the hard part. The hard part is that “average number employed” has at least four legitimate UK definitions, and the answer changes depending on which you pick, so the definition has to travel with the number. Publish the period, the population, the leaver rule and the denominator method alongside the percentage, or it cannot be compared with anything, including your own figure from last year.
The formula, and what the CIPD actually says
The standard method has a name. In its Resourcing and Talent Planning report, the CIPD sets out what it calls the “crude wastage” method:
Labour turnover = (Number of leavers in a set period / Average number employed in the same period) x 100
Its definition of a leaver is more inclusive than most people assume. Leavers are employees who “for any reason, have left the organisation on a permanent basis, and include those taking up employment elsewhere, retirees, redundant employees and those dismissed. It does not include internal transfers.”
A retirement counts. A redundancy counts. A dismissal counts. A promotion from kitchen porter to chef de partie does not. If you have been quietly excluding redundancies because they were a management decision rather than a resignation, your rate is not the crude rate and should not be compared with anything that is.
The CIPD also publishes the caveat that people quoting the formula leave out. Readers, it says, “should be aware that this method has some shortcomings. For example, it takes no account of the characteristics of the workforce or the length of service of the leaver.” That is the professional body for HR saying the headline number hides who left and how long they had been there. Everything below is built on that admission.
Calculate voluntary turnover separately as well. The CIPD’s retention guidance distinguishes employee-initiated departures from employer-initiated ones such as redundancy and dismissal, and the split matters because you can influence one far more than the other.
The denominator is the real decision
There is no single UK standard for the average number employed. Four conventions are in live official use, and they disagree.
| Whose method | The denominator | Where it comes from |
|---|---|---|
| CIPD | ”Average number employed in the same period”, method unspecified | Resourcing and Talent Planning report |
| Cabinet Office | Average of opening and closing headcount, or more frequent data where held | Turnover in the Civil Service: guidance on calculations |
| Companies Act 2006, s.411 | Sum of each month’s headcount divided by the number of months | The average employee number in your statutory accounts |
| NHS Scotland | No average at all: staff employed at the start of the period | NES official workforce statistics |
The Cabinet Office guidance is the most useful of the four for a shift employer. Its method is “the number of leavers within that period divided by the average of staff in post over the period”, and while it permits the opening-plus-closing average, it adds that where organisations “have access to more frequent data on the number of staff in post, these figures can be used to calculate the average staff in post”. You almost certainly hold monthly headcount. Use it.
Here is why. Take a seasonal business reporting on a calendar year, with 100 staff in January and 100 in December, but 200 through July and August. Thirty people leave across the year. On the opening-plus-closing average the denominator is 100 and the rate is 30%. On a monthly average that captures the peak the denominator might be 140, and the same thirty leavers give 21%. Nobody behaved differently. One method simply cannot see your summer.
The distortion runs the other way too: an employer that grew from 80 to 160 gets a flattering rate from a closing-headcount denominator and a harsh one from an opening-headcount denominator. If your business is stable, opening-plus-closing is fine and the Cabinet Office sanctions it. If it swings, take the monthly average. A rate can also legitimately exceed 100% where the same post churns more than once in a year, which is common in high-volume hourly roles and is not an arithmetic error.
The Companies Act route has a practical bonus. Section 411 already requires the notes to your accounts to disclose the average number of persons employed, found by adding each month’s headcount and dividing by the number of months. Your finance team produces that number every year, so using it means the rate you take to the board reconciles to the statutory accounts.
Headcount or full-time equivalent, and never a mix
Both are defensible and they answer different questions. Headcount tells you about people churn: how many individuals you had to replace, induct and train. Full-time equivalent tells you about capacity loss: how much resourced time walked out of the door. NHS England publishes turnover both ways side by side, which is the cleanest evidence that neither is wrong; Skills for Care calculates on directly employed staff and reserves FTE for sizing the workforce.
The rule is not to mix them, because a headcount leaver count over an FTE denominator means nothing. Where people hold more than one contract, decide early whether you are counting people or posts. Skills for Care counts filled posts, so its post count exceeds its people count; NHS England warns that headcount totals may not equal the sum of their components because some staff hold multiple roles. If a bank nurse works two contracts at your group and resigns one, you need to have decided in advance whether that is a leaver.
Who counts as a leaver when staff are casual, bank or agency
This is where a shift workforce breaks the textbook. Each of these has a settled legal position, even though almost no statistical authority spells out the turnover treatment.
| Situation | The legal position | Defensible treatment |
|---|---|---|
| Casual and zero-hours staff | Under ERA 1996 s.212, continuity survives weeks of absence caused by “a temporary cessation of work” | A gap between shifts is not a termination. Count a documented end to the relationship, not a quiet month |
| Bank and pool workers | Skills for Care excludes bank, pool and indirectly employed staff from its turnover denominator | Keep a separate direct-employee rate and a bank inactivity measure |
| Agency workers | Under the Agency Workers Regulations 2010 reg 3, the worker contracts with the agency, not the hirer | An assignment ending at the hirer is not a leaver from the hirer. Track assignment churn separately |
| Fixed-term and seasonal ends | Non-renewal of a fixed-term contract “is considered to be a dismissal” (gov.uk) | It counts. Split planned seasonal expiry from avoidable departures |
| Internal transfers and promotions | The CIPD’s crude wastage definition excludes them | Out of the organisation-wide rate. Show them as site or department outflows where useful |
| TUPE transfers | Under TUPE 2006 reg 4 the transfer does not terminate the contract, and Acas confirms continuity and start date are preserved | Not a leaver, even though the outgoing payroll shows a mass leaver event |
| Death in service | HMRC requires the date of death in the FPS leaving date and no P45 | It lands in payroll-derived turnover unless you exclude it. Keep it out of voluntary turnover |
| Retirement | Acas confirms there is no default retirement age and an employer generally cannot force retirement | The CIPD counts retirees as leavers. They are worker-initiated, so classify separately rather than treating them as unavoidable |
The Cabinet Office does something worth copying. It runs two turnover measures off one dataset: internal moves between departments are excluded from Civil-Service-wide turnover but included in departmental turnover. For a multi-site group that is exactly right. A chef moving from your Leeds site to your Manchester site is a leaver for the Leeds general manager and is not a leaver for the group.
The four dates that all claim to be the leaving date
Turnover looks like an HR number until you reconcile it with payroll, at which point one person can carry four legitimate dates. The last day worked is what HMRC’s Full Payment Submission field asks for. The effective date of termination under ERA 1996 s.97 is the date notice expires, which can be later. The payroll leaving date is whenever it was keyed in, and it only reaches HMRC on the run that pays them. The P45 issue date is not a leaving date at all.
Four traps follow, and each produces a real discrepancy between two departments looking at the same month.
Leavers arrive late. Someone only enters HMRC’s data on the run that pays them, so a last shift in late March paid in April lands in April. And gov.uk instructs that where someone leaves on or after 6 April you “do not put their leaving date in the same FPS as their final payment”, reporting them instead in the first submission of the new tax year with zero pay. That creates a visible April cluster of leavers HR recorded in March.
Absence of pay is not absence of employment. HMRC’s guidance provides an irregular payment pattern indicator, set to “Yes” if the employee “is not being paid regularly (for example, they’re a casual employee or on long-term sick leave) or if you’re not going to pay them for 3 months or more”. With that flag set, casual and zero-hours staff stay live on the payroll with no leaving date through long gaps between shifts. Without it, a quiet quarter can read as a departure. For an employer whose winter rota is a third of its summer rota, this one field explains a lot of phantom turnover.
Re-hires double-count by design, because HMRC requires a different payroll ID when you re-employ someone, restarting year-to-date information from £0.00 within the same tax year. A seasonal worker who returns every summer is two payroll records. And wrong dates are sticky: HMRC’s correction guidance says to update your own records but warns against reporting the amendment in your next FPS because it “may create a duplicate record for the employee”, so your corrected copy and HMRC’s can diverge permanently.
The fix is unglamorous. Pick the HR effective termination date as the turnover event, store the other three alongside it, and reconcile HRIS, payroll and rota records monthly rather than at year end. A P45 should corroborate a leaver, not define who counts as one.
The metrics that show what the headline rate hides
The CIPD conceded that the crude rate ignores length of service. These put it back.
The stability index is the standard companion, and it has three live UK definitions, so state which you used. The CIPD divides staff with a year’s service or more by the total in post one year ago, times 100. Invest Northern Ireland’s official business guidance uses current total employees as the denominator instead. NHS England defines it as the percentage of the opening cohort still in post at period end, which makes in-year hires who leave in-year invisible. Same workforce, three answers.
A survival curve is the most useful thing a shift employer can plot, and Invest NI describes the method plainly: plot the number of employees leaving against their length of service. Turnover is usually highest among new hires and falls the longer people stay. Where the curve drops steepest is where your money is going.
Tenure bands turn that into something you can act on. Skills for Care’s analysis of care worker turnover between March 2024 and March 2025 found 34.0% for those in role under a year, falling through 26.5%, 22.0% and 17.3% to 15.3% for the longest-serving group. One employer type, and a spread of nearly nineteen points that a single headline rate erases.
Retention is not automatically 100 minus turnover. That identity holds only when both measures use the same cohort, period and denominator, which they usually do not.
What to compare your number against
The most quoted UK figure is the CIPD’s 34%, from its analysis of the ONS Annual Population Survey covering January 2022 to December 2023. It splits into 27.4% who moved to a new employer and 6.6% who were not working a year later, and by industry runs from 25% in public administration and defence to 52.2% in accommodation and food services.
Those are not employer turnover rates. It is a household survey tracking whether individual workers changed employer between two points a year apart, not leavers divided by your headcount. Comparing your internal crude rate against 34% compares two different quantities that happen to share a percentage sign.
For an employer-reported comparator, Brightmine’s survey of UK organisations put median voluntary turnover at 10.3% and median total turnover at 14.0% for 2024. Those are medians of what employers reported themselves, the same species of number as yours. The full method sits behind a subscriber wall, so treat the base and definition as unconfirmed, which is exactly the caution this guide asks you to apply to every borrowed figure.
Adult social care has unusually good published data, and it is unusually easy to misquote. Skills for Care reported turnover among directly employed staff in England at 23.1% for 2024/25, roughly 335,000 leavers, with the independent sector at 24.7%, local authorities at 12.5% and direct payment recipients at 16.8%. By service it was 25.5% residential and 22.2% domiciliary; by role, 29.7% for care workers and 32.8% for registered nurses. A separate Skills for Care publication reported the independent sector at 23.6% for 2025/26. Do not average those. They cover different years, populations and scopes, and Skills for Care warns its figures are not directly comparable across editions.
The scope note matters as much as the rate. Skills for Care measures “directly employed staff only (permanent and temporary staff). Leavers from agency roles, for example, aren’t included”, and counts only establishments still operating. Benchmark an all-in rate that includes your agency churn against 23.1% and you will look worse than you are. Roughly half of what social care employers record as leavers are moves to another employer inside the sector, since 53% of recruitment comes from within adult social care.
Three rules keep benchmarking honest: match the population, match the period, match the denominator. Where you cannot, say so. Your own figure from twelve months ago, computed the same way, beats any external comparator.
What a leaver actually costs
The cost figures circulating on this topic are worse than the rate figures. Two get quoted constantly, and they measure different things in different decades.
The famous one is £30,614, from Oxford Economics research for Unum published in February 2014. Read the report and it stops being usable. It covers five sectors only: IT and technology, accounting, legal, media and advertising, and retail. It looks solely at employees earning £25,000 or more, at a time when the report itself notes only 15% of retail workers earned above that, and hospitality, adult social care and agency labour were not studied. The headline splits into £5,433 of logistical cost and £25,181 of lost output, and that second figure includes £12,054 of “lost capital income”, an economist’s estimate of forgone profits rather than cash your business pays out. It is a twelve-year-old white-collar number, and applying it to a £12-an-hour shift worker is indefensible.
The CIPD’s median cost per hire is current and much narrower: £1,500 for most employees and £2,000 for senior managers and directors, from April 2024 fieldwork. It covers in-house resourcing time, advertising and agency or search fees, and nothing else. No cover, no induction, no lost output. It is a recruitment invoice, not a turnover cost, and the bases are small, 207 and 246 respondents from a survey of 1,016.
The most useful recent UK work is the Joseph Rowntree Foundation’s February 2026 study, built bottom-up from the financial records of 11 mainly independent care providers. Its medians per new care worker hired were £800 recruitment, £884 training, £3,683.86 in agency cost per departing worker and £3,288.13 for ramp-up, assuming a new starter works at 50% efficiency while training. The median total financial cost came to £7,870.38, with lost output reported separately at £9,282.13. JRF is explicit that the components do not sum to the total and that the two headline figures overlap, so adding them double-counts. Eleven self-selected providers is not a representative sample either.
Nobody else’s number fits your business. Build your own, and keep three totals apart. Cost per hire is what you spend to fill the role: advertising, agency and search fees, DBS and right-to-work checks, interviewing time at real salaries, HR admin. Direct replacement cost adds what a shift employer actually bleeds cash on and the published studies mostly miss: overtime premium to cover the gap, agency cover at the markup, induction and training hours, the trainer’s and manager’s time at loaded cost, and kit, uniform or licences. Economic cost adds evidenced output lost while the post is empty and while the new starter ramps up, less the salary and on-costs you did not pay during the vacancy and the output the cover actually delivered.
Only 31% of UK organisations that track their turnover data calculate what it costs them, rising to 43% of larger organisations and falling to 18% of SMEs. Doing it from your own invoices is what makes it defensible.
Do you have to report any of this?
For most UK employers, no. There is no general statutory duty to calculate or publish a staff turnover rate. Gender pay gap regulations bite at 250 or more employees but prescribe six figures about pay, bonuses and quartiles, none concerning turnover, and neither Companies Act section 172(1) statements nor the 2024 Corporate Governance Code prescribes a workforce metric.
Adult social care is the exception. CQC’s Provider Information Return asks how many staff have left the service in the past 12 months, providers get four weeks to respond, and CQC states that a service which does not complete and return it by the deadline will not be rated better than “requires improvement” for well-led. That is the closest thing in the UK to a statutory turnover return. The separate monthly Capacity Tracker submission, which is a hard duty, carries no leavers field at all.
What to do with the number
A single company-wide rate is a number you can report. A segmented one is a number you can act on.
Skills for Care’s data is the best UK template for the cuts that pay off, because it shows how wide the spread inside one employer type can be. Care workers on zero-hours contracts had 30.6% turnover against 22.9% for those on more than 35 contracted hours a week. Where the registered manager had been in post under a year it was 28.7%, against 21.3% where the manager had twenty years or more. Workers with five positive employment factors in place, relatively higher pay, guaranteed hours, training, a relevant qualification and full-time work, showed 14.4% turnover against 42.2% for those with none.
Read those as associations, because that is what they are. Skills for Care says so directly: the analysis “shows the association between each variable and the turnover rate; this is not necessarily the cause or the reason for workers leaving”. Anyone selling you a retention product on the back of numbers like these, ours included, should be held to the same standard.
Cut your own rate by site, line manager, tenure band, contract type and shift pattern. Calculate monthly, report a rolling twelve months quarterly, and publish raw counts next to the percentages, because on a ten-person site one extra leaver moves the rate ten points and means almost nothing. NHS England takes this seriously enough to suppress rates where organisational change has broken the denominator, and warns that its method “may create unusually high rates where small numbers of staff are involved”. Before attributing a jump to a manager, check whether payroll, rota or coding practice changed. Then set a baseline and leave the method alone, because the most common reason a turnover trend looks dramatic is that somebody improved the calculation halfway through.
On pay timing specifically, we will not tell you it fixes turnover. Nest Insight’s independent review found the evidence on earned-wage access “remains patchy” and noted that the positive evidence is largely generated by product providers, and its own study did not measure turnover at all. The DWP’s 2024 employer survey found 4% of GB employers offering earned-wage access, 17% offering some form of advance and 73% offering neither. No UK study we can find establishes that earlier access to earned pay causes lower turnover. What is dependable is narrower: a cash squeeze between paydays is one reason hourly staff leave, and if you remove it without charging them for the privilege, you remove that reason. Measure it the way this guide describes and judge it on your own numbers.
For the wider retention picture, our guide to reducing staff turnover in hospitality covers the levers alongside pay timing, and the employer case sits on earned-wage access for employers. Sector views are on hospitality and care.
Frequently asked questions
What is the formula for staff turnover?
Divide the number of leavers in a period by the average number of people employed over that period, then multiply by 100. The CIPD calls this the crude wastage method. Leavers include everyone who left permanently, so resignations, retirements, redundancies and dismissals all count, but internal transfers and promotions do not.
How do you work out the average number employed?
There is no single UK standard. The Cabinet Office permits the average of opening and closing headcount but says that where you hold more frequent headcount data you should use it instead. Companies Act 2006 s.411 defines a monthly average, which is the figure your statutory accounts already carry. For a seasonal workforce use a monthly average, because an opening-plus-closing average cannot see your peak.
What is a normal staff turnover rate in the UK?
It depends on the measure. Brightmine’s survey of UK organisations put median employer-reported turnover at 14.0% total and 10.3% voluntary for 2024. The widely quoted 34% from the CIPD is a different quantity: it comes from the ONS Annual Population Survey and tracks whether individual workers changed employer over a year, not leavers divided by an employer’s headcount. Sector matters enormously, from 25% in public administration to 52.2% in accommodation and food services on that population measure.
Do zero-hours and agency staff count as leavers?
Agency workers contract with the agency rather than with you, so an assignment ending is not a leaver from your payroll. For casual and zero-hours staff, ERA 1996 s.212 preserves continuity through a temporary cessation of work, so a gap between shifts is not a termination. Count a documented end to the relationship instead, and set HMRC’s irregular payment pattern indicator so a quiet period does not read as a departure.
Does a TUPE transfer count as staff turnover?
No. Under TUPE 2006 reg 4 a relevant transfer does not terminate the contract of employment, and Acas confirms length of service and contractual rights carry across unchanged. The outgoing employer’s payroll will still show a mass leaver event, which is why turnover computed from payroll alone spikes falsely after a transfer.
How much does staff turnover cost?
Calculate it from your own records rather than borrowing a figure. The two most quoted UK numbers measure different things: the CIPD’s median cost per hire is £1,500 for most employees and £2,000 for senior roles, covering recruitment spend only, while the £30,614 figure from Oxford Economics dates to February 2014, covers five white-collar sectors and only employees earning £25,000 or more, and includes an imputed lost-capital-income element that is not cash you pay. Keep cost per hire, direct replacement cost and full economic cost separate, and never add figures that overlap.
Is a UK employer legally required to report staff turnover?
There is no general statutory duty to calculate or publish a turnover rate. Gender pay gap reporting applies at 250 or more employees but concerns pay, not turnover, and neither the Companies Act s.172(1) statement nor the UK Corporate Governance Code prescribes a workforce turnover metric. Adult social care is the exception: CQC’s Provider Information Return asks how many staff left in the past 12 months, and a service that misses the deadline cannot be rated better than “requires improvement” for well-led.
What is the difference between turnover and the stability index?
Turnover counts who left. The stability index measures how well you hold on to experienced people, and it has three live UK definitions, so state which you used. The CIPD divides staff with a year’s service or more by the total in post a year ago; Invest Northern Ireland uses current total employees; NHS England measures the opening cohort still in post at period end, which makes in-year joiners who leave in-year invisible. Retention is not simply 100 minus turnover.
Further reading and sources
- CIPD, Resourcing and talent planning report 2020: the crude wastage formula, the leaver definition and the CIPD’s own caveat about length of service.
- Cabinet Office, Turnover in the Civil Service: guidance on calculations (December 2019): the average-staff-in-post method and the instruction to use more frequent headcount data where held.
- Companies Act 2006, s.411: the statutory monthly-average definition of the average number employed.
- GOV.UK, employee leaving (PAYE) and payroll information to report to HMRC: leaving dates, the 6 April rule, re-hire payroll IDs and the irregular payment pattern indicator.
- CIPD, Benchmarking employee turnover (June 2024): the 34% UK churn figure and the industry spread, measured on ONS Annual Population Survey data.
- Skills for Care, The state of the adult social care sector and workforce in England 2025 (October 2025): sector turnover rates, the scope exclusions and the segmentation data.
- Joseph Rowntree Foundation, The hidden cost of low pay in the social care sector (February 2026): a bottom-up UK cost model built from provider financial records.
- Oxford Economics for Unum, The Cost of Brain Drain (February 2014): the origin of the £30,614 figure, and the methodology that limits where it applies.
- CQC, Provider Information Return: the leavers question and the well-led consequence of missing the deadline.
If you are working out whether pay timing belongs in your retention plan, how earned-wage access works sets out the mechanism and free earned-wage access explains the funding model behind it.