Employee Attrition Rate Calculator
From beginning and ending headcount and separations, compute the employee attrition rate.
輸入資料
計算結果
重點速覽：Attrition rate = separations / average headcount x 100%, where average headcount = (beginning + ending) / 2. Using the average smooths within-period changes. Example: beginning 200, ending 180, 30 left gives average 190 and 15.79%. WARNING: Distinguish voluntary vs involuntary and regrettable vs healthy attrition; compare with industry benchmarks; any decision must comply with Hong Kong's Employment Ordinance.
計算公式
平均員工數 = (期初員工數 + 期末員工數) ÷ 2。
流失率 = 期內離職人數 ÷ 平均員工數 × 100%。
$$\overline{N} = \dfrac{N_{\text{begin}} + N_{\text{end}}}{2}$$$$\text{Attrition Rate} = \dfrac{\text{Separations}}{\overline{N}} \times 100\%$$使用說明
- Enter the beginning and ending headcount.
- Enter the number of separations in the period.
- View the average headcount and the attrition rate.
以期初 200 人、期末 180 人 (平均 190 人) 為例，不同離職人數對應的流失率對照表
| 期初人數 | 期末人數 | 平均員工數 | 期內離職 | 流失率 |
|---|---|---|---|---|
| 200 人 | 180 人 | 190 人 | 20 人 | 10.53% |
| 200 人 | 180 人 | 190 人 | 30 人 | 15.79% |
| 200 人 | 180 人 | 190 人 | 40 人 | 21.05% |
| 220 人 | 200 人 | 210 人 | 42 人 | 20.00% |
| 150 人 | 150 人 | 150 人 | 30 人 | 20.00% |
平均員工數 = (期初 + 期末) ÷ 2;流失率 = 離職人數 ÷ 平均員工數 × 100%。注意末兩列:期初期末人數不同、離職數不同，但流失率同為 20% — 說明比較流失率時，分母的計算口徑必須一致才有意義。
理財情境案例
擴張中的公司：為何要用平均員工數作分母
某公司期初 250 人、期末 230 人，期內共 36 人離職。平均員工數 = (250 + 230) ÷ 2 = 240 人，流失率 = 36 ÷ 240 × 100% = 15.00%。
假如改用期末的 230 人作分母，流失率會被算成 36 ÷ 230 ≈ 15.65%;若用期初的 250 人，則為 36 ÷ 250 = 14.40%。三個數字都不同 — 這正是為什麼要用『平均員工數』:它平滑了期內人數的變動，讓流失率更能代表整段期間的一般人力規模，跨期比較也更公平。若公司人數在期內大幅波動 (如季節性擴張)，甚至可改用每月月底人數的平均，但關鍵始終是口徑前後一致。(數字為便於說明的假設值。)
拆分自願與非自願：同樣 16.67%，意義大不同
某公司平均員工數 240 人，期內 40 人離職，總流失率 = 40 ÷ 240 × 100% ≈ 16.67%。乍看偏高，但把 40 人拆開:其中 28 人是主動辭職 (自願)、12 人是公司裁員或辭退 (非自願)。
自願離職率 = 28 ÷ 240 ≈ 11.67%，非自願離職率 = 12 ÷ 240 = 5.00%。這個拆分至關重要:高比例的『自願離職』(尤其若集中在高績效者或關鍵部門) 才是留才的警號，往往指向薪酬、管理、文化或發展機會的問題;而『非自願離職』反映的是經營決策與績效管理，成因與對策完全不同。若這 40 人裡多數是公司主動精簡的低績效者，16.67% 反而可能是健康的新陳代謝。這說明:同一個總流失率，拆開來看才知道是危機還是常態。
常見問題
Why use average headcount as the denominator?
Headcount changes within the period, so either endpoint alone can distort the rate. Averaging (beginning + ending) / 2 fairly reflects the typical size over the period. For very volatile headcount, use the average of month-end counts, but keep the basis consistent.
What attrition rate is high?
No universal standard — it varies by industry, role and region (retail/F&B higher; professional services lower). Compare with your own history and peers, and break down by department and level. Build a baseline (e.g. your 3-year average) and see if the current figure is flat, improving or worsening.
Do all departures count?
For the total rate, yes. But for management, split voluntary vs involuntary (layoffs, dismissal) and regrettable (high performers) vs healthy. Total turnover reflects filing load; voluntary rate signals pay/management issues; regrettable loss is the real warning.
Is attrition rate the same as turnover rate?
Both translate to 'employee attrition/turnover rate' and share the same formula (separations / average headcount). Usage sometimes differs (turnover = all exits; attrition = natural exits not backfilled), but not universally. What matters is defining the numerator and denominator clearly and keeping them consistent. Our turnover-rate calculator takes average headcount directly; this one derives it from beginning/ending counts.
What Hong Kong compliance points apply?
The rate itself is just a metric, but decisions from it are governed by law. Redundancies/terminations must follow the Employment Ordinance (notice or pay in lieu, earned leave pay, and severance payment or long-service payment where eligible, not both). Handling employee data must comply with the Personal Data (Privacy) Ordinance. Avoid discrimination under Hong Kong's anti-discrimination ordinances when analysing groups. This tool is for calculation only, not legal advice — consult Labour Department guidance and professionals.
相關工具
參考資料
內容審核：香港計算器財經團隊。人力資源指標之計算邏輯與公式屬通用管理會計常識，人事決策相關規定參考香港勞工處《僱傭條例》及個人資料私隱專員公署指引，結果僅供參考，實際以相關機構公佈為準。