How to Calculate Attrition Rate by Department

Attrition rate by department uses the exact same formula as company-wide attrition. What the number reveals once you stop averaging it across the whole organization is that a healthy-looking company average can be hiding a department that's bleeding people, or confirming that the problem really is everywhere.

Attrition rate versus turnover

Attrition and turnover get used interchangeably in casual conversation, but they mean different things. Attrition counts employees who leave and are not immediately replaced, which means the workforce shrinks. Turnover counts departures where the role gets backfilled, so headcount stays roughly stable even as people cycle through it. Retention is the mirror image of both, the share of people who stayed, and tracking it alongside attrition tells you more than either figure alone.

Not all attrition is the same kind of signal, either. Voluntary attrition means someone chose to leave, through resignation or retirement. Involuntary attrition is company-initiated: restructuring, layoffs, performance-based terminations. SHRM's 2025 data puts the median involuntary rate at just 3% SHRM 2025. Internal transfers don't count as departures at all in the formula, since the person hasn't left the company, though they still shrink a department's working capacity in the short term. Demographic-concentrated attrition, meaning departures clustered by gender, age, or ethnicity, deserves its own separate line of tracking rather than getting folded into the general number. Retirement attrition is predictable in aggregate, and tracking its acceleration feeds directly into succession planning.

The reason to sort these out before calculating anything is simple: what belongs in the numerator changes depending on which question you're asking. A department calculation that lumps involuntary terminations in with voluntary resignations tells you volume but tells you nothing about cause. Temporary workers, contractors, employees on leave, and internal transfers should all be excluded from the count entirely; none of them represent the kind of departure the formula is built to capture.

The core formula, applied step by step to a department

The math itself doesn't change when you move from company level to department level. Attrition Rate equals the number of employees who left, divided by the average number of employees, multiplied by 100. Average headcount is the start-of-period count plus the end-of-period count, divided by two, which accounts for growth or shrinkage across the window instead of anchoring to a single snapshot.

First, pick a measurement period, monthly, quarterly, or annual, since each one serves a different planning purpose. Second, pull the department's headcount at the start of that period. Third, pull headcount at the end. Fourth, average the two. Fifth, count every permanent departure from that department during the window, resignations, retirements, and eliminated positions that weren't backfilled, while excluding anyone who transferred internally to another team. Sixth, multiply by 100 to express as a percentage.

A company starts the year at 250 employees, ends at 230, and records 32 departures across the period SHRM 2025. Average headcount comes out to 240, and the attrition rate is 13.33% SHRM 2025. Swap in a single department's numbers and the calculation runs identically, no adjustment needed.

Don't multiply a monthly rate by 12 to estimate an annual figure. That overstates the real risk, because it ignores compounding and seasonal variation in when people actually leave. If you want an annual number, run the annual formula, total departures over the full 12 months against the full-year average headcount, rather than extrapolating from a single month's snapshot.

Why company-level calculation obscures key visibility

A company-wide figure is an average, and averages are built to smooth things out. That's exactly the problem. Mercer's 2025 survey of 2,617 organizations puts the U.S. voluntary attrition average at roughly 13%, a number that reads as unremarkable on a dashboard Mercer 2025 US Turnover Survey BLS JOLTS Pave. But if a Sales department inside that same company is running at 24%, the aggregate isn't reassuring, it's concealing an active retention crisis one function at a time Mercer 2025 US Turnover Survey BLS JOLTS Pave.

Concentration affects how confidently you can pinpoint the cause. If most of a company's departures are coming out of one function, the cause is almost certainly local, a manager, a workload pattern, a compensation gap specific to that role, rather than something company-wide. That distinction changes what gets fixed and who fixes it.

Volume alone also misses who is leaving. A department that loses its highest performers, its most specialized technical staff, or the people holding the closest client relationships takes damage that has nothing to do with the raw percentage. Gallup's 2024 Employee Retention and Attraction research found that 52% of U.S. employees are actively watching for or seeking a new job, the highest self-reported turnover risk Gallup has measured since 2015 riseworks.io. With that much latent risk sitting in the workforce, knowing exactly which departments are most exposed is a basic operating requirement, not just a nice-to-have riseworks.io. Department-level tracking is what converts attrition from a number you report after the fact into a signal you can act on before the damage compounds.

Department-level benchmarks to compare your numbers against

Benchmarks exist to give a number context, not to hand you a target. A rate that's perfectly healthy in one function would set off alarms in another, so the comparison only means something when it's function to function.

Pave's 2024 to 2025 U.S. data breaks this out clearly. Marketing and Sales is at 24% annual turnover BLS JOLTS Pave. Human Resources runs at 23%: the function responsible for designing retention strategy is itself experiencing some of the highest attrition in the company SHRM 2025 Pave. Product is at 23% as well, partly a function of how differently product roles get scoped from one company to the next, which creates more lateral movement between employers SHRM 2025 Pave. Engineering sits lowest at 17%, and Pave's Great Stay research attributes part of that to fear of AI displacement driving a kind of job-hugging behavior among technical staff through 2024 and 2025.

European tech tells a similar story with different numbers. Ravio's 2025 data puts average attrition across European tech at 17.4%. Engineering again comes in lowest, at 12%, consistent with the U.S. pattern Ravio. The UK average is 19%, above the European average and representing an 11% increase from 2024 Ravio.

Industry matters just as much as function. Mercer's 2025 survey shows voluntary attrition in retail and wholesale running at 26.7%, against just 8.2% in insurance, a gap wide enough to make clear that a single company-wide benchmark rarely applies across sectors.

Below 10% is the healthy range for most companies, and lower is simply better. Between 10% and 15% is manageable but worth watching. Between 15% and 20% is elevated and suggests something systemic is going on Eddy and Paychex research. Above 20% is critical and calls for immediate intervention Eddy and Paychex research. The practical move is to run the department formula, then check the result against the function-specific benchmark rather than the company average: a Sales team at 20% looks fine against a generic company-wide number, but it's actually below its own peer benchmark, while a Sales team at 28% is in crisis territory by any standard Eddy and Paychex research.

How to segment further once you have department-level numbers

A department-level rate is a starting point. It's not a diagnosis, and treating it as one is where a lot of HR analysis stalls out Gallup EY 2025 Cost Update.

Tenure is the first cut worth making. Departures within the first 90 days usually point to a hiring or onboarding failure, not a department culture problem, and the fix for one looks nothing like the fix for the other. Role level is the second cut: individual contributors leaving at a higher rate than managers often signals a stalled growth path, while managers leaving predicts future IC attrition down the line.

Departure type changes the interpretation just as sharply. Splitting voluntary from involuntary departures within a department can flip the story, since high involuntary attrition in one team might point to a performance management gap rather than a retention failure. Demographic concentration deserves its own look too, if departures cluster by gender, age, or ethnicity, the raw rate isn't explaining anything on its own and the pattern needs a separate investigation. And performance quartile might be the most important cut of all: losing low performers can actually be a healthy sign, while losing top performers is regrettable attrition, the figure that matters most for whatever the business is trying to keep running.

None of this segmentation is academic. Gallup's 2024 research found that 42% of employees who voluntarily left said their manager or organization could have done something to prevent it. That statistic reframes the entire exercise: attrition isn't some inevitable cost of doing business, it's a measurable governance problem, and segmentation is the tool that turns a vague number into a specific answer about whether the issue is the department, the manager, the role design, or a compensation gap that just happens to surface in one team before it shows up anywhere else.

Measurement cadence for tracking changes over time

Monthly calculation catches sudden spikes fast, especially in departments with enough headcount to make the percentage meaningful, though it gets noisy in smaller teams. Quarterly calculation smooths out that noise and works well for mid-year reviews and board reporting. Annual calculation is the standard for benchmark comparison, and it should always run as the full-year formula, total departures across 12 months against the full-year average headcount, rather than a monthly figure stretched out by multiplication.

Small departments need a specific caveat attached to them.

Trend direction carries as much weight as the current level does. A department at 18% that was at 25% six months ago is heading in the right direction, even though 18% still sounds high in isolation MIT study. A department at 14% that was at 8% six months prior is a warning sign, even though 14% looks fine on its own MIT study. The number without its history is only half the story.

Building this out doesn't require anything elaborate. A simple tracking table by department, period, starting headcount, departures split by voluntary and involuntary, ending headcount, average headcount, and the resulting rate, updated every period, is enough to make patterns visible over time. Coefficient.io notes that a basic spreadsheet with built-in fields is a common starting point before teams eventually move the tracking into an HRIS, and there's nothing wrong with starting there. In a small department of 8–12 people, a single departure produces an attrition rate that looks alarming by any benchmark, so the absolute headcount should be noted alongside the percentage so the number isn't misread.

What to do when a department's number is high

The rate itself tells you volume and concentration. It does not tell you cause, and treating it as if it does is the most common misstep in reading this data. Cause requires qualitative work layered on top of the arithmetic.

The first question to ask is whether the elevated attrition is voluntary or involuntary, since the interventions for each are entirely different. If voluntary attrition is running high in one department, a few causes recur constantly. Manager effectiveness is one, since the department is the reporting unit but the manager is frequently the actual variable. Compensation competitiveness relative to the market for that specific function is another; the gap between Marketing and Sales at 24% and Engineering at 17% partly reflects how differently those labor markets price talent BLS JOLTS Pave. McKinsey research cited in Pin's 2026 guide found that lack of advancement beats pay as the top reason people quit. And workload distribution deserves attention, because when a department loses people and the roles don't get backfilled, the remaining staff absorb the extra load, which accelerates the next wave of departures, a pattern sometimes described as trickle-down burnout.

Exit interviews, stay interviews, and engagement surveys are the instruments that explain what the rate surfaces. The number tells you where to look. It doesn't tell you what to change. And it certainly doesn't tell you whether the attrition is regrettable, that requires knowing who left, not just the count, which is why performance quartile has to travel alongside the rate rather than getting calculated separately and forgotten.

Stacey Staaterman, CEO of Staaterman Strategic Advisory, put it directly in a 2026 piece for HiBob: "There is a strong relationship between retention and organizational change… Being curious about the WHYs behind attrition and resistance is job #1, from there, an organization can deploy solutions". Left unaddressed, high department-level attrition carries costs well beyond the immediate gap on an org chart: lost institutional knowledge, eroding team culture, and mounting pressure on whoever's left standing. Gallup estimates the cost of replacing a single employee at somewhere between half and twice their annual salary, and puts the collective cost of voluntary departures to U.S. businesses at roughly $1 trillion a year.

Automated workforce infrastructure and the operational cost of tracking and acting on attrition data

The formula itself is simple enough for a spreadsheet. The real challenge at scale is data quality: headcount figures that don't reconcile between systems, departure records that lag behind what actually happened, department assignments nobody's updated since the last reorg.

Manual HR and payroll processes make this worse in ways that compound quietly SHRM 2025. EY's 2025 Cost Update study puts the average cost of a single manual HR data entry at $4.86, with complex tasks like benefits enrollment running as high as $89.00 per employee when there's no automation involved Gallup EY 2025 Cost Update. For a company of 500 people, combined manual HR processing commonly exceeds $100,000 a year in direct labor alone, before factoring in the cost of errors Gallup EY 2025 Cost Update. Payroll errors, which occur in 20% of manually processed payrolls per research cited by Eddy and Paychex, are not just financial problems (they erode trust and accelerate voluntary attrition) Eddy and Paychex research. The attrition problem and the operational problem turn out to be the same problem wearing different labels.

AI adoption inside HR functions is moving fast, from 26% of organizations in 2024 up to 43% in 2025, according to Landbase research SHRM 2025 Conference of State Bank Supervisors (CSBS) landbase.com. But adoption alone hasn't translated into results: a cited MIT study found only 5% of organizations saw a measurable return on their AI investment, largely because deployments got bolted onto existing workflows piecemeal instead of getting built into how the work actually happens SHRM 2025 Conference of State Bank Supervisors (CSBS) landbase.com.

When payroll, headcount records, benefits, and offboarding all live inside the same system, attrition data is accurate by default, with no reconciliation lag, no misclassified departures, and no manual data entry creating the errors that undermine the very metrics HR is trying to use. An AI-native employee management platform that owns payroll, compliance, benefits, and offboarding end to end keeps the headcount and departure data feeding the attrition formula current at all times, which is what turns the metric from something retrospective into something a department can actually act on before the next quarter's number confirms what everyone already suspected.