Regretted attrition is the share of employees who left a workforce that the organisation would have wanted to keep. This page is for People and Finance leaders at 200–2,000-person scale-ups who already have a headline employee attrition rate and need to know whether they are losing people they would have kept — not for university dropout rates, research-study attrition, or machine-learning datasets.
It shows how to calculate an employee attrition rate, monthly or annually, how employers use “attrition” and “turnover”, and how to split leavers into regretted (or regrettable) and non-regretted attrition using a written rule that starts with a calibrated performance rating. A clearly labelled worked example runs through the arithmetic. The page does not offer a “good” or “healthy” attrition percentage; there is no single official target.
Disclosure: Ash Taherian is Product Sales Executive at Revolut People, which sells performance management software. Treat product mentions as coming from an interested party.
What employee attrition is, and how to calculate the rate
Employee attrition is the share of people who left a workforce during a period. The usual way to express it is:
Employee attrition rate = (leavers in the period ÷ average headcount in the same period) × 100
Average headcount = (opening headcount + closing headcount) ÷ 2
Use the same formula for a month or a year, changing only the period you count. Do not invent a special “annualisation” shortcut — for example, multiplying a monthly rate by 12 — unless you can cite a named method you are prepared to stand behind. This page does not use one. The result is a method you can repeat, not a published industry average.
Count people who left the organisation in the period, and be explicit about who is in and out of the headcount: employees in the workforce, not contractors you have chosen to exclude, and not internal moves that stay inside the company. Keep voluntary resignations, involuntary or company-chosen exits, and internal moves in separate columns from the start, even if you also report one combined rate.
Illustrative example, not a benchmark. A scale-up starts the year with 500 employees and ends it with 520. Average headcount is (500 + 520) ÷ 2 = 510. If 76 people left during the year, the annual employee attrition rate is 76 ÷ 510 × 100 = 14.9%. In a month that opened at 510 and closed at 512, with 6 leavers, average headcount is (510 + 512) ÷ 2 = 511, and the monthly rate is 6 ÷ 511 × 100 = 1.2%. These figures are fictional, and they are not a target.
Attrition rate vs turnover rate
Employers use the words in two ways, and neither is a single “correct” usage. Some treat attrition rate and turnover rate as the same percentage: leavers in the period, divided by average headcount, times 100. Others reserve “attrition” for roles that are not replaced and “turnover” for all exits, or use turnover for the replacement cycle — the leaving, the vacancy, and the hire.
What matters is that you pick one definition, write it down, and use it every time you report the number. If one team counts only resignations and another counts every exit, the two percentages cannot be compared. The same applies to monthly and annual figures: they are the same formula on a different period, not interchangeable numbers.
Keep voluntary, involuntary and internal moves in separate columns so you can explain a movement. A rise driven by a managed-performance or redundancy programme is a different conversation from a rise in resignations. Internal moves are not leavers if the person is still in the organisation; record them, but do not fold them into the attrition rate unless that is the definition you have written down.
What regretted attrition is
Regretted attrition — also called regrettable attrition, and sometimes regretted turnover — is the subset of exits, usually voluntary, that the organisation would have wanted to prevent. The judgement is not “we liked them” after the person has gone. It is a written rule, agreed before anyone leaves, and applied when the resignation arrives.
The Playbook starting point is the last calibrated performance rating against a published scale. Use the employee performance rating scales guide for the scale itself, and performance calibration — including how to run a calibration meeting — so the rating is a shared, evidenced decision, not a manager’s gut. Critical-role or succession status can be an additional criterion; see succession planning for People teams. A solid performer in a hard-to-replace role can be a regretted exit without being in the top box.
Non-regretted attrition is the other side of the same split: exits the company chose, or would not work to reverse. That often includes many involuntary or managed-performance exits. A poorly run process is not automatically “healthy” attrition. When the exit sits inside a formal performance process, use the UK capability procedure guide and the PIP template for UK managers — do not treat every formal process as good attrition.
Tag the exit close to the resignation, with People and the manager using the written rule together. Do not wait until a year-end clean-up, when memory and leftover feeling have already rewritten the story. The same published scale should be used for every leaver in the period, including people who were not rated in the last cycle: record that fact, and decide in the written rule how unrated exits are handled, rather than improvising case by case.
How to calculate regretted attrition
Two figures are both useful, and they answer different questions:
Regretted attrition rate = (regretted leavers ÷ average headcount) × 100
This is a rate you can track over time, on the same denominator as the headline attrition rate.
Share of voluntary exits that were regretted = (regretted leavers ÷ voluntary leavers) × 100
This shows what share of resignations you wanted to keep.
A third, optional figure is a high-performer attrition rate: voluntary leavers in the top rating groups, divided by average headcount in those groups, using the same published scale. It is not the same as regretted attrition, because a regretted exit can sit outside the top boxes. Use it as a companion number, not a substitute.
The table below continues the same fictional company as the formula section, and adds a prior year so you can see how a stable headline can hide a change in the mix. Every number is an illustration, not a benchmark or a target.
| Measure | Year 1 | Year 2 |
|---|---|---|
| Opening headcount | 490 | 500 |
| Closing headcount | 510 | 520 |
| Average headcount | 500 | 510 |
| All leavers | 70 | 76 |
| Headline employee attrition rate | 14.0% | 14.9% |
| Voluntary leavers | 50 | 60 |
| Regretted leavers | 8 | 18 |
| Regretted attrition rate | 1.6% | 3.5% |
| Share of voluntary exits that were regretted | 16% | 30% |
| Voluntary leavers in the top rating groups | 6 | 12 |
| Average headcount in those groups | 120 | 125 |
| High-performer attrition rate | 5.0% | 9.6% |
The year 2 arithmetic is the same as the opening example: (500 + 520) ÷ 2 = 510, and 76 ÷ 510 × 100 = 14.9%. The regretted attrition rate is 18 ÷ 510 × 100 = 3.5%. The share of voluntary exits that were regretted is 18 ÷ 60 × 100 = 30%. The high-performer attrition rate is 12 ÷ 125 × 100 = 9.6%.
Year 1 uses the same method: average headcount (490 + 510) ÷ 2 = 500; headline rate 70 ÷ 500 × 100 = 14.0%; regretted rate 8 ÷ 500 × 100 = 1.6%; share of voluntary exits 8 ÷ 50 × 100 = 16%; high-performer rate 6 ÷ 120 × 100 = 5.0%.
The headline barely moved, from 14.0% to 14.9%. The regretted attrition rate more than doubled, and the share of resignations the company wanted to keep rose from 16% to 30%. If the board had only seen the combined percentage, the year would have looked uneventful.
How to read the number against performance
A rise in headline attrition driven by non-regretted, company-chosen exits is a different problem from a rise in exits among people rated above the bar. Before you brief the board, look at the split by rating, by team and by tenure. A company-wide percentage hides all three.
Connect the regretted split to talent density, without redoing that calculation here. Losing people who were above the bar is what moves talent density hardest. The Playbook already sets out the density formula, and how a high-performer leaver moves the number, in how to measure talent density. Use the free talent density calculator to see the effect of those leavers on your own figures. This page is the attrition-rate how-to and the regretted / non-regretted split; it does not replace those two.
If a regretted leaver sat on a succession slate, the next question is critical-role coverage and who is ready now — that work belongs on the succession planning page. If the last 1:1s were thin or cancelled, the warning often showed up there first; keep a regular record with the one-to-one meeting template so a regretted exit is less of a surprise. The wider process these ratings sit in is the performance management framework.
There is no “good” or “bad” attrition percentage on this page, and this is not a retention-tactics list. A short list of what to look at next is enough:
- The regretted and non-regretted split for the last 12 months, by rating, team and tenure, using the written rule.
- Whether anyone who left sat on a critical-role or ready-now slate.
- Whether 1:1 notes were in place in the months before the resignation.
- Whether pay conversations are outstanding — that is a separate discussion, using a merit matrix, not a catch-all “retention strategy”.
The free Playbook tools are there if you want to run the density or review-process numbers alongside this split. The next practical step is to apply the split to last year’s leaver list, with names removed, and last-cycle ratings if you have them.
Common mistakes
- Reporting only a company-wide rate. A stable headline can hide a rise in regretted exits, as the worked example shows.
- Tagging “regretted” from memory after the person has gone. Agree the rule in writing before anyone leaves, and apply it at the point of resignation, starting from the last calibrated rating.
- Mixing involuntary exits into the regretted rate. Company-chosen exits belong in the non-regretted (or at least a separate) column unless your written rule says otherwise, and you apply it consistently.
- Comparing monthly and annual figures as if they were the same. They use the same formula on a different period. Do not treat 1.2% in a month as comparable with 14.9% in a year.
- Treating an academic or vendor “average attrition” as your target. This page does not use one. Read your split against your own history.
- Using an engagement “attrition risk” score as if it were actual attrition. Sentiment about staying is useful context. It is not a count of leavers, and it is not the regretted-attrition calculation below.
Where Revolut People fits
Revolut People does not, on its public documentation, calculate an attrition rate, a regretted-attrition rate, or a high-performer leaver rate, and it does not flag exits as regretted. Do not read this section as a built-in regretted-attrition report.
What the product does offer on this subject is Attrition Risk, an Engagement Survey metric. It groups responses to questions tagged “Attrition risk” into three sentiment levels: Low risk (employees indicate they expect to stay), Medium risk (some uncertainty about staying), and High risk (employees indicate they are less likely to stay). Two templates include an Attrition Risk question by default: the Generic Engagement Survey (20-question) and the Deep-Dive Engagement Survey (60-question). The tag can be added or removed on a question. An example question on the docs page is: “Do you see yourself working at our company for the foreseeable future?”
Attrition Risk is an indicator of employee sentiment. It is not a prediction or guarantee that an employee or group will leave. Use it as a signal, with other engagement results, not as a substitute for the actual regretted-attrition calculation on this page, and not in isolation to take action on a named person. AI Insights on Attrition Risk, if you use them, are available after the survey round ends and are based on the final response set; they do not calculate regretted attrition.
Source: Revolut People, “Attrition Risk” (engagement survey documentation), checked 15 November 2026.
If you want a second look at the split, book a 15-minute review with Ash, Product Sales Executive. Bring last year’s leaver list with names removed and, if you have them, last-cycle ratings. It is a fit-check, not a sales pitch, and not a trial.
FAQ
What is employee attrition, and how do you calculate an attrition rate?
Employee attrition is the share of people who left a workforce during a period. For a chosen period, divide the number of people who left by the average headcount in that period, then multiply by 100. Average headcount is usually opening headcount plus closing headcount, divided by two. Use the same formula for a month or a year, on that period’s figures. This is a method, not a published industry average, and there is no single official “good” attrition rate.
Is attrition the same as turnover?
Not reliably. Some employers use attrition rate and turnover rate for the same percentage. Others reserve “attrition” for roles that are not replaced and “turnover” for all exits, or use turnover for the replacement cycle. Pick one definition, write it down, and keep voluntary, involuntary and internal moves in separate columns so you can compare one period with the next.
What is regretted (or regrettable) attrition, and how do you calculate it?
Regretted attrition — also called regrettable attrition, or sometimes regretted turnover — is the subset of exits the organisation would have wanted to prevent, judged against written criteria agreed before anyone leaves. Start with the last calibrated performance rating, not a manager’s view after the person has gone. The regretted attrition rate is regretted leavers divided by average headcount, times 100. A second figure — regretted leavers divided by voluntary leavers — shows what share of resignations you wanted to keep.
How do you split leavers against performance ratings?
Tag each exit close to the resignation, using a written rule that starts with the last calibrated rating against a published performance rating scale. People and the manager apply that rule together. You can then report regretted and non-regretted rates, and a high-performer attrition rate: voluntary leavers in the top rating groups divided by average headcount in those groups. Do not invent the tag months later from memory.
What is non-regretted attrition?
Non-regretted attrition is exits the company chose or would not work to reverse. That often includes many involuntary or managed-performance exits. A poorly run process is not automatically “healthy” attrition; see the UK capability procedure guide and the PIP template for how a formal process should be handled. Keep these exits in a separate column from regretted resignations.
Is regretted attrition the same as high-performer attrition?
Not always. High-performer attrition looks only at voluntary leavers in the top rating groups, as a share of average headcount in those groups. A solid performer in a hard-to-replace or critical role can still be a regretted exit without being in the top box. Use both figures, and use succession planning when the person sat on a critical-role slate.
Is there a good or healthy attrition rate?
No single official “good”, “bad” or “healthy” employee attrition or regretted-attrition percentage is used on this page. A stable headline rate can still hide a rise in regretted exits. Read the split by rating, team and tenure against your own history, not against a vendor average.
Can an engagement “attrition risk” score replace this calculation?
No. In Revolut People, Attrition Risk is an Engagement Survey metric — an indicator of employee sentiment, not a count of leavers and not a prediction that someone will leave. Use it as a signal, with other engagement results, not as a substitute for the regretted-attrition calculation on this page, and not in isolation to take action on a named person.
Related guides
How to measure talent density · talent density calculator · employee performance rating scales · performance calibration · how to run a calibration meeting · succession planning · one-to-one meeting template · UK capability procedure · PIP template for UK managers · performance management framework · merit matrix · free Playbook tools