Free · 33-page guide, no email required
The cost of finding out too late.
What burnout-driven attrition actually costs an Indian company, built from the bottom up, line by line, in rupees, with every assumption written down so it survives the meeting where someone asks where the number came from.
Written for CHROs, heads of people and founders who already know burnout is expensive and need to prove it to a finance team. It is a working model, not a report: nine inputs, eight cost lines, a worked example for a 200-person company, and a one-page business case template.
The same model also comes as a live Excel workbook. Change the blue cells to your own headcount, salaries and notice periods, and the cost per exit, the annual exposure and the break-even all recalculate. The guide explains the method; the spreadsheet is the thing that ends up circulating.
Read the guide
33-page PDF: the model explained, with a worked example and every source dated.
Download the PDFNo email, no form. Edition one, October 2026.
Why most retention business cases fail.
Not because the costs are small. Because the case arrives built on a statistic nobody can defend. The most quoted figure in the field is that replacing someone costs half to twice their annual salary. It comes from a 2019 Gallup article that publishes no method, no sample and no cost categories, while peer-reviewed estimates of the same quantity sit near 21% and 23.5% of salary.
That is a ten-fold spread. Quote the high end and a numerate CFO can halve it twice and still be inside the literature. Quote the low end and you have argued yourself out of your own case. This guide replaces the multiplier with arithmetic you can show, and the band it produces, roughly 0.3× to 1.6× of cost to company, falls out as a result rather than an assumption.
The eight lines the model costs.
Each one is defined, with a formula, a default range and the double-counting trap that sits next to it. Seven are recovered only by preventing the exit. The eighth is recovered by noticing earlier, which is a different argument entirely.
L1
Separation and handover
The notice period you pay in full and receive a fraction of, plus the colleagues’ hours spent receiving the handover. In India this line is unusually large, because notice periods run 30 to 90 days.
L2
Vacancy and coverage
Output not produced while the seat is empty, and the seat stays empty far longer than time-to-fill suggests, because the person you hire has to serve their own notice first.
L3
Recruitment
Agency fees at 8.33% to 16.67% of CTC, sourcing, recruiter time, and the interview-panel hours almost nobody costs: about 32 hours of senior technical time for one engineering hire.
L4
Onboarding and ramp
The largest line in most builds, and the one where published benchmarks fall apart. The guide gives you a three-point estimation method instead of a borrowed statistic.
L5
Manager and team drag
Fifty to eighty manager hours across a quarter, plus the work absorbed by colleagues, which is the mechanism by which one exit becomes two.
L6
Second-order attrition
The exits that follow an exit. Included as a scenario with the uplift you choose, because no published India coefficient exists and inventing one would discredit the rest.
L7
Pre-exit decay
The months of reduced output before the resignation. The only line early detection recovers even on the exits you fail to prevent, and so the only one that does not rest on a counterfactual.
L8
Role-specific exposure
Quota, named accounts, key-person risk. Stated beside the total and never inside it, because deferred bookings are not lost revenue and a CFO will say so.
What it refuses to do.
The exclusion list is the most persuasive page in the document. It is also the reason the rest of it is believable.
- No invented benchmarks. Where we could not trace a figure to a primary source, it is excluded and the omission is explained, including the time-to-productivity benchmarks everyone quotes and nobody can source.
- No customer stories. The worked example is a clearly labelled construct. It is not a client, anonymised or otherwise.
- No unmeasurable costs in the totals. Employer-brand damage, lost institutional memory, morale and work never attempted are all real and all absent from every number, because pricing them would mean making them up.
- No misused statistics. The WHO’s “$4 returned per $1” figure is about scaling up treatment access, not workplace wellbeing spending. The guide says so, and lists every source with the claim it does not support.
What is inside.
- The number nobody has, and the five sources that carry the weight in India.
- Why “half to twice salary” will not survive your CFO.
- The model: eight lines, with formulae and the double-counting traps.
- Presenteeism and absenteeism, priced at population level.
- A worked example: 200 people, two roles costed in full, the annual aggregate.
- What early detection is worth: break-even, sensitivity, and the value of lead time.
- The one-page business case, and the four questions your CFO will ask.
- How you will know whether it worked, before you spend anything.
- A blank assumptions log to fill in and attach.
- Every source, dated, with its limits stated.
The workbook carries the same model as live formulas across eight sheets: inputs, the ramp curve, cost per exit, annual exposure, break-even with a sensitivity grid, the assumptions log, and the sources.
Read next
- Employee retention is a measurement problem before it is a perks problemWhy retention budgets get spent on the wrong things, what actually predicts an exit, and how to see attrition risk by department months before anyone resigns.
- The attrition rate formula, and what your number actually meansEvery attrition rate formula in one place, worked through with real headcount, plus a calculator and an honest read on whether your number is high.
- Burnout is the leading indicator. Attrition is the receipt.How burnout becomes a resignation, why exit interviews arrive too late to be useful, and what a leading indicator has to do before you can act on it.