Employee Turnover vs Attrition: The Difference and How to Calculate Each

“Our attrition is 15%.” “No, turnover is 15%, attrition is lower.” I have heard this argument in management meetings more than once, and usually both people were right — they simply meant different things by the same words.

In 21 years in HR I have built turnover reports in a bank, a TV group, a game developer and a software company. The formula was never the hard part. The hard part was agreeing on who counts. This guide separates the two terms, gives you the formula for each, shows how to classify every person who left, and explains what each number means for the business.

If you already have your list of leavers, go straight to the free leaver classifier.

Both words describe people leaving the company. The difference is what happens to the seat after the person walks out.

• Turnover — a person leaves and you hire someone into the same position. The headcount stays the same; the company pays for the search, onboarding and the months before the newcomer is fully productive.
• Attrition — a person leaves and the position is not refilled. The role is eliminated, the work is redistributed, a hiring freeze is on, or someone retires and the company decides it no longer needs that job. The headcount goes down.

Be honest about one thing: the market does not use these words consistently. In my own course decks the lesson is called “Attrition” and the formula on the slide is called “turnover rate” — and that is normal, because in everyday analytics most people use the two as synonyms. You will also meet a third usage, where “attrition” means only voluntary resignations.

So the first rule is not a formula. Write down your definition in the calculation method section of the report and keep it the same from period to period. Then anyone in the company can recalculate your number and get the same result.

This is the total number: everyone who left, voluntarily or not, replaced or not. For a period shorter than a year, annualize it: the rate for the period × 12 ÷ number of months. I cover the formula, the monthly and annual versions, the stability index and the cost of turnover in detail in a separate guide: Employee Turnover Rate: formula, calculator and how to read the numbers. Here we go one step further and split that total.

Use exactly the same denominator as for total turnover. Then the parts add up: turnover = replaced turnover + attrition, and nobody in the meeting can claim the numbers contradict each other.

Example. A company starts the year with 200 people and ends it with 190. Average headcount is 195. Thirty people left during the year. Eleven of their positions were closed: seven roles were eliminated in a restructuring, and four more were not refilled after resignations — the work went to colleagues.

• Turnover: 30 ÷ 195 = 15.4%
• Replaced turnover: 19 ÷ 195 = 9.7%
• Attrition: 11 ÷ 195 = 5.6%

One practical rule that saves arguments: count a departure as attrition only when the position is closed in the staffing plan. An open vacancy that recruiting has not filled yet is still a replacement — it is simply late. Otherwise your attrition grows every time recruiting is slow, and falls again when they catch up.

Replaced or not replaced is the business question. The second question is why the person left, and here I use the classification from my HR Analytics course:

1. Voluntary — the employee decided.
   • Voluntary avoidable — the company could have influenced the departure: pay, manager, workload, career.
   • Voluntary unavoidable — something beyond the company’s control, such as moving to another country.
2. Involuntary — the company decided: dismissal for performance or conduct, layoffs, eliminated roles.

Across both groups there is one more split: regret or no regret. Regret means the person’s performance was above average; no regret means it was below average.

Each type is calculated the same way:

Turnover by type = leavers of that type ÷ average headcount

In the HR Metrics course I show it like this: a company of 3,000 people where 30 left voluntarily and 30 involuntarily has 1% voluntary and 1% involuntary turnover. Same total, two completely different conversations.

Why bother? Because the overall number can worry you, and when you take it apart you find it is not so bad. Involuntary departures tell you about recruiting quality and performance management, not about whether people want to stay. Unavoidable departures you will never prevent. What remains — voluntary, avoidable and regretted — is the part retention work can actually move.

In my AI People Analytics course I go one level deeper and sort the reasons for leaving into eight classes: pay gap, career stall, manager toxicity, burnout, wrong hire (early departure, a selection error), org change, life event, and forced exit. Two dimensions sit on top of them:

• Push or pull. Push factors drive a person out: pay below market, no promotion for years, a bad manager, overload. Pull factors attract them elsewhere: an external offer, relocation, a new market. HR mainly influences push — remove the push cause and attrition goes down.
• Voluntary or involuntary. Forced exits always go into a separate pile. If you mix them with resignations, the picture of why people leave is distorted.

This is also where the classifier rows above come from: avoidable departures are mostly push, unavoidable ones are mostly life events, and involuntary ones are a different analysis altogether.

Look at that case through the classification. Almost a third of the departures were the bank’s own decision — that part is a question for hiring and performance management, not retention. Among the reasons for leaving, family circumstances are mostly unavoidable. “Found a higher position” is avoidable — and a career model answers it directly. The 432 promotions are internal moves: they do not count as turnover for the bank at all, even though people left the agent role.

One more note on the number itself: 21.7% is for six months. Annualized at the same pace (× 12 ÷ 6), it is about 43% — compare it with full-year figures only after that step.

Replaced turnover is a cost line. Every replacement is a vacancy, a search, onboarding and a newcomer who needs time to reach full output. When half of your recruiting is replacements, the company is running to stand still. This number belongs in the conversation with the CFO about retention budgets.

Attrition is a capacity line. When seats close, the work either disappears or lands on the people who stayed. That can be a deliberate choice: a company that needs to reduce headcount may decide not to refill some seats instead of laying people off. But unplanned attrition — seats closed because nobody could be found, or because the budget was cut quietly — shows up later as overload, burnout and the next wave of resignations. Track it next to workload, not only next to the hiring plan.

Attrition is also something you forecast. In the AI People Analytics course we build the forecast as the sum of individual departure probabilities by risk zone, which gives an expected number of departures by department for the quarter. That number feeds workforce planning: how many seats will free up, and which of them you actually need to refill.

Retention rate is the other side. It is the percentage of employees who started and completed the period in a segment. Transfers into and out of the segment are excluded. It helps when one high-churn role inflates the turnover of a whole department.

The company-wide number is an average of very different situations. Cut both turnover and attrition by department, job title, gender, age, tenure, whose initiative, reason, regret, salary level, engagement, loyalty and parental status.

The rule that people break most often: divide the leavers of a segment by the headcount of that same segment, not by the total headcount. Women who left ÷ women working. People with 10–15 years of tenure who left ÷ people working with 10–15 years of tenure. Split both the leavers and the people who stayed in the same way, otherwise the segments are not comparable.

And do not compare raw rates between groups that are built differently. Interchangeable positions — cashiers, customer service representatives, sales representatives — have a higher average turnover rate than specialised or technical roles. Tenure and location shift the rate as much as anything you control.

Classification starts with an honest answer to “why did you leave?” — use the exit interview template to collect it. The full analytics path — turnover by type, segmentation, retention rate, the turnover report outline and the retention budget — is in my HR Analytics Course: People Analytics Training Online. If your main task is bringing the numbers down, start with the Employee Retention Course: Reduce Staff Turnover.