Three people in a meeting say "we lost 15% of the team last year," and all three mean something different. One is talking about turnover, one about attrition, and one is quietly doing retention math in reverse. The turnover rate vs. attrition rate distinction (and retention rate, and the SaaS refugee "churn") isn't pedantry — the four words are four lenses on the same departures, and they answer different questions. Here's the short version, then the formulas, then one 100-person company run through every metric so you can see where they diverge.
Turnover Rate vs. Attrition Rate vs. Retention Rate: The Short Answer
- Turnover counts every departure during a period, regardless of reason, as a share of average headcount. It assumes the seat gets refilled (or could be). It's the number benchmarks are built on.
- Attrition is the subset of departures that aren't backfilled — retirements, eliminated roles, a hiring freeze that lets seats go empty. It's how a company gets smaller without a layoff.
- Retention flips the lens: of the people here at the start, what share are still here at the end? It follows a cohort forward instead of counting exits.
- Churn is borrowed from subscription software, where it means customers cancelling. Applied to people it's just turnover — usually the voluntary kind — and "employee churn" is safe to read as a synonym.
The convention isn't universal — plenty of teams use "attrition" and "turnover" interchangeably. If your leadership uses both words in the same slide deck, pin down which one you mean before anyone sets a target. The definitions above are the most common usage.
The Formulas, Side by Side
Turnover rate
The same formula as in how to calculate employee turnover rate:
Turnover rate = (Number of separations during the period ÷ Average number of employees during the period) × 100
Where average headcount is (Beginning headcount + Ending headcount) ÷ 2. Separations means everyone who left, for any reason. Internal transfers and promotions never count.
Attrition rate
Attrition rate = (Departures not backfilled during the period ÷ Average number of employees during the period) × 100
Some organizations instead compute attrition as net headcount reduction: (Beginning headcount − Ending headcount) ÷ Beginning headcount × 100. The two agree when hiring is only backfilling and diverge once you open new roles, so know which one your HRIS reports.
Retention rate
Retention rate = (Employees at the start of the period who are still employed at the end ÷ Employees at the start of the period) × 100
Note the denominator: starting headcount, not the average, and the numerator ignores anyone hired during the period. Retention is a cohort metric, which is why it refuses to add up to 100% with turnover. More on that below.
One Company, Three Metrics
A company starts the year with 100 employees. During the year, 15 people leave. It replaces 12 of them and lets 3 seats go: two retirements it didn't need to refill and one role it eliminated. Assume for now that all 15 leavers were on the January payroll. Ending headcount is 100 − 15 + 12 = 97.
Turnover rate. Average headcount is (100 + 97) ÷ 2 = 98.5. Separations are 15. So (15 ÷ 98.5) × 100 = 15.2%. (Dividing by starting headcount gives exactly 15% — close here, but the gap widens when headcount moves fast.)
Attrition rate. Departures not backfilled: 3. So (3 ÷ 98.5) × 100 = 3.0%. The net-headcount version, (100 − 97) ÷ 100 × 100, also gives 3%, because every hire this year was a replacement.
Retention rate. Of the 100 people on payroll in January, 85 are still there in December. (85 ÷ 100) × 100 = 85%.
Same year, same 15 goodbyes: 15.2% turnover, 3% attrition, 85% retention. None of them is wrong. Turnover measures how much hiring you did to stand still. Attrition says the company got smaller. Retention says how well you held onto the specific people you had.
Voluntary, Involuntary, and Regrettable
Turnover has one more split that matters more than any of the above. Voluntary turnover is resignations and retirements — people who chose to go. Involuntary turnover is layoffs and terminations — decisions the company made. Same formula, filtered numerator. If 11 of our 15 resigned, 2 retired, and 2 were let go, that's 13 voluntary (13.2%) and 2 involuntary (2.0%).
Inside voluntary sits the slice almost nobody tracks and everybody should: regrettable turnover, the people you'd have fought to keep. If 6 of those 11 resignations were strong performers, your regrettable rate is about 6%, and that's the only number where "as low as possible" is unambiguously the goal. A poor fit who leaves on their own is the system working. Tag every departure at exit; it takes thirty seconds, and since the Work Institute has reported that roughly 3 in 4 voluntary departures are preventable, the regrettable rate is the one that points at what you can fix.
Why Retention Rate + Turnover Rate ≠ 100%
In the clean example above, 85% retention and 15.2% turnover sum to 100.2%, close enough that people assume they're mirror images. They aren't. The small reason is denominators: retention divides by starting headcount (100), turnover by average headcount (98.5), so any headcount change drives a wedge between them.
The big reason is new hires who leave. Change one assumption: 2 of the 15 departures were people hired in March who quit in September. They're separations, so turnover still counts all 15 (15 ÷ 99.5 ≈ 15.1%, the two extra hires nudging the average). But they were never in the January cohort, so retention only loses 13 of the original 100: 87%. Now retention plus turnover is 102%, and the extra 2 points are exactly your first-year churn. A company with a revolving door in its first six months can post a healthy retention rate and an ugly turnover rate at once. Both are true, about different people.
How to Annualize a Monthly Rate
Monthly turnover numbers are small and easy to shrug off, so annualize before you judge. There are two ways, and they don't agree.
The simple annualized rate: Annualized turnover ≈ Monthly turnover rate × 12. A 1.5% month is roughly an 18% year. It's a linear approximation that slightly overstates the true compounded figure, but for management reporting it errs on the side of taking the problem seriously.
The actual annual rate: the year's total separations divided by the year's average headcount (or, approximately, the twelve monthly rates added up).
Why they differ: ×12 assumes this month is every month. A 200-person company that loses 3 people in January (1.5%) projects to 18%; if the year ends with 24 departures on an average headcount of 200, the real rate was 12%. January was just a bad month. Summing monthly rates also drifts from the true annual figure whenever headcount moves, because each month uses its own denominator. Use ×12 for early warning and the measured figure for benchmarking, and never present the projection as the measurement. The turnover rate formula in Excel lays out monthly, rolling-12-month, and YTD versions; our employee turnover rate calculator does the annualization for you.
Which Metric to Report, and Which to Manage
Report turnover to leadership, annual, split voluntary and involuntary. It's the number that maps to benchmarks and to cost. Show attrition alongside it whenever headcount is deliberately shrinking, so a planned reduction isn't read as a retention crisis.
Manage regrettable voluntary turnover, monthly, by team and tenure band. That's where the preventable departures and the levers are. Retention rate is best used for cohorts — how much of last year's hiring class is still here, how much of engineering survived the reorg. A great answer to a specific question, a mushy headline metric.
Benchmarks, Briefly
Compare whatever you land on to your industry, not to a generic figure. A good turnover rate is commonly cited as somewhere around 10% a year, but the variance between sectors is wider than the rule of thumb itself. US Bureau of Labor Statistics JOLTS data puts total separations across all industries somewhere in the high 30s to mid 40s percent annually, from roughly 70–80% in accommodation and food service down to 18–25% in government. Those are total separations; an internal voluntary-quits rate typically runs roughly half to two-thirds of that, so match definitions before you compare. The full table is in employee turnover rates by industry, and the longer discussion of what counts as healthy is in what is a good employee turnover rate.
From Definitions to Dollars
Once you know which number you're looking at, price it. Replacement cost estimates cluster in a consistent band: roughly 33% of salary as a conservative floor (Work Institute), 50–60% as a typical figure (SHRM), and one-half to two times salary at the high end (Gallup). Run the mid-range on our example company: 100 employees × $65,000 average salary × 15% turnover × 50% replacement cost = $487,500 a year. Attrition is notably cheaper — the 3 seats you didn't refill carry no recruiting or ramp bill, which is exactly why the two metrics deserve separate lines. The cost of employee turnover calculator runs this on your own numbers, and what employee turnover really costs itemizes the bill.
Since the expensive slice is voluntary and mostly preventable, the cheapest lever is worth naming. Gallup and Workhuman found employees who feel inadequately recognized are about twice as likely to say they'll quit within a year; Deloitte/Bersin links strong recognition cultures to up to 31% lower voluntary turnover. Against a $487,500 problem, 31% is roughly $150,000 a year — arithmetic we walk through in recognition vs. turnover: the retention math.
Full disclosure, this next bit is ours. Propsly is a Slack-native peer recognition tool: anyone types /props @maria +20 for shipping the migration #teamwork, the give lands in a feed channel like #wins, and the leaderboard does the rest. Free for unlimited users at 200 props per person per month; the $50/month flat Pro tier adds analytics and automated monthly gift-card rewards. It won't change how you define attrition, but it's one of the few cheap things that moves the voluntary number.
So: turnover is everyone who left, attrition is the seats you didn't refill, retention is who stayed from the cohort you started with, and churn is turnover in a SaaS costume. Report the first, manage the regrettable slice of it, and stop expecting retention and turnover to add up to 100 — they were never measuring the same people.