Attrition Cost Calculator

Expected Attrition Loss (EAL) for any employee, any level, any currency, any latent cost.

Sample values are loaded so you can see how it works. Replace them with your own data, or use Settings & data → Clear all values.

Employee

Probability helper (from your history)

Method

Each factor is a % of CTC. Rename, add or remove factors freely.

Template percentages are only starting points. Calibrate them to your own measured costs.

Extra latent costs

Client loss, penalties, team morale dip, rehire fee, etc. Do not repeat anything already counted above.

Result

–
Expected loss = EAL × probability–
Break-even retention spend–
Group exposure (all leave)–
Group expected loss–

Saved scenarios

Save several employees or roles from the Calculator tab to compare them and total the expected loss.

Input toolkit: how to arrive at each figure

General rules: (1) take figures from records, not memory; (2) for time, percentage and count inputs use the median of the last 5 to 10 similar cases; (3) note the source and date; (4) if no record exists, give your best estimate and tag the group Estimated; (5) enter 0 if it does not apply; (6) never count the same cost twice. All examples are illustrative.

How to use

  1. Settings & data: set currency and number format.
  2. Employee: enter fully loaded annual CTC. Get the probability of leaving from the helper (leavers ÷ average headcount of the same band or function over 12 months), or type your own.
  3. Method: use Quick for screening. Use Bottom-up for decisions: every field names its record source and how to measure it.
  4. Tag evidence for each cost group as Measured (from a record such as payroll, ATS, invoice, contract or log), Estimated (a stated basis from a manager or Finance) or Assumed (no data). The result shows what share of EAL is measured.
  5. Enter 0 for any component that does not apply to the role.
  6. Read the result: EAL is the cost if the person leaves. Expected loss = EAL × probability is the break-even ceiling for retention spend.
  7. Save scenarios to compare roles. Export JSON to keep your model.

Formulas

Hourly cost = CTC ÷ annual working hours
Replacement = CTC×fee% + sourcing + interview hrs×hourly + onboarding
Vacancy = (value ÷ working days)×days vacant + interim cover
Ramp-up = (value ÷ 12)×months×(1 − productivity%) + training + manager hrs×hourly
Core = Replacement + Vacancy + Ramp-up + Exit
Knowledge & team = extra defects×rework hrs×hourly + handover hrs×hourly
  + team size×(CTC÷12)×dip%×dip months + peers at risk×P(follow-on exit)×Core
  P(follow-on exit) = exits followed by another exit ÷ total exits
Business impact = margin×P(impact)×loss%×months at risk÷12
  + min(penalty/day×delay days, cap)×P(delay) + ΔARO×incident cost×window
  P(impact) = past exits that caused impact ÷ total exits; window = (vacancy days + ramp days) ÷ 365
EAL = Core + Knowledge & team + Business impact + latent items
Expected loss = EAL × P(leave) Group = result × headcount

Making the figures objective

  • Every probability is derived from counts in your own history, never typed as a hunch.
  • Use contribution margin, not revenue, for revenue at risk. Use the contract for penalties and the risk register for incident costs.
  • For defect rate, dip % and months, compare data from the 3 months before and after 10 to 20 past exits and use the median.
  • Where judgment remains (ΔARO, dip %), tag the group Estimated and write down the basis outside the tool.
  • Do not count one cost twice, for example a penalty in Business impact and again as a latent item.
  • Track the measured share over time and raise it by replacing estimates with records.

Data owners

DataOwner
CTC, notice, exit settlements, attrition historyHRIS, Payroll, HR
Time-to-fill, agency fees, sourcing spend, interviewer hoursTalent Acquisition
Margin, value per role, invoices, incident costFinance, Risk
Training cost, time-to-productivityL&D, line manager
Onboarding and equipment costIT, Admin
Defect, output and ticket dataDelivery, QA, line manager
Penalty clauses, severance termsLegal, contracts
Account and project impact historyAccount and project owners

Limits

This is a one-year, undiscounted expected-value model and a decision aid, not an audited figure. Test it against several real past exits before relying on it.

Currency & format

en-IN shows lakh / crore. Other locales show K / M / B. Enter all amounts as full numbers in your currency.

Data

Your model is also remembered in this browser, on this device only.