Attrition is discussed in almost every BPO operations review, usually as a percentage. Percentages are easy to get used to. A rupee figure tends to get more attention, and putting one on attrition shows why it deserves the effort. This article builds the cost for one process step by step, starting with what is easy to see and moving to what is usually missed.
Step 1: how many people actually leave
A voice process has an average headcount of 200 agents and a monthly attrition rate of 7%. Monthly attrition is normally calculated as leavers in the month divided by average headcount for the month, where average headcount is often taken as the mean of opening and closing headcount. At 7%, about 14 agents leave each month, or 168 over a year if the rate holds steady.
Multiplying 7% by 12 gives an “annualised” attrition of 84%. That is a common way to report it, but it does not mean 84% of today’s agents will be gone in a year. Some seats churn several times while long-tenured agents stay put. What matters for costing is the number of replacements needed, and that is 168.
Definitions differ by company: some exclude involuntary exits (terminations for performance or conduct), some exclude trainees who leave during classroom training and track them separately. Each choice changes the percentage, but every leaver still has to be replaced, so for cost purposes it is safest to count everyone.
Step 2: the direct cost per replacement
The visible costs of replacing one agent include sourcing (job portals, referral bonuses, consultant fees, walk-in drives), assessment and interviews, background verification, onboarding, and the salary paid during classroom training when the new joiner is not yet productive. Suppose these add up to ₹35,000 per hire for this process.
168 replacements × ₹35,000 = ₹58.8 lakh a year. This is the number most finance teams can calculate, and it is already significant for a 200-seat process.
Step 3: the learning curve
A new agent who finishes training is not as productive as the experienced agent who left. Handle times are longer, more calls need help from a team leader, and the first-contact resolution rate is lower. A simple way to price this is to estimate productivity relative to a tenured agent for each early month.
Say a new joiner is at 60% of full productivity in the first month on the floor and 80% in the second, reaching full productivity by the third. The lost output is 0.4 + 0.2 = 0.6 agent-months. If an agent costs about ₹30,000 a month, that is ₹18,000 of paid capacity that did not produce full output. Across 168 hires, it adds ₹30.2 lakh a year.
Total so far: ₹58.8 lakh + ₹30.2 lakh = about ₹89 lakh, or roughly ₹53,000 per leaver. The learning-curve cost alone adds about half again to the direct cost.
Step 4: the costs that are harder to count
Several further effects are real but difficult to put a precise number on. It is better to name them than to invent figures.
- Buffer headcount. To keep enough trained agents on the floor, the process carries extra people in the training pipeline at all times. Under most FTE contracts, these are not billable.
- Quality and SLA impact. New agents make more errors and escalate more. If service credits apply when SLAs are missed, attrition can reduce revenue as well as raise cost.
- Load on the people who stay. Team leaders spend more time on side-by-side support and less on coaching tenured agents. Experienced agents absorb the harder calls. Both can push more people towards leaving, which is why attrition tends to feed on itself.
- Customer experience. Repeat calls caused by incorrect first answers cost the client money and patience.
Step 5: what a two-point improvement is worth
If monthly attrition falls from 7% to 5%, annual leavers drop from 168 to 120. Using the countable cost of about ₹53,000 per leaver, that is 48 fewer replacements and a saving of about ₹25.4 lakh a year, before counting any of the harder-to-measure effects.
This figure is the budget argument for retention work. If a mentoring programme, a better shift-allocation policy or a fixed-shift option for long-tenured agents costs ₹10 lakh a year and plausibly delivers that improvement, it pays for itself. Without a rupee figure, such proposals tend to be treated as “nice to have”.
Where the money leaks most: early attrition
Not all leavers cost the same. An agent who leaves after two years has repaid their training many times over. An agent who leaves in week three of training has consumed most of the hiring cost and produced nothing. In many BPOs, a large share of total attrition happens in the first 90 days, so splitting the attrition figure by tenure band is the single most useful analysis.
Early attrition usually points to something before or at the start of the job: a role that was oversold during hiring, shift timings that were not made clear, a candidate profile poorly matched to the process (for example, strong conversational English required but not tested), or a rough first week on the floor without enough support. These causes are more fixable than general market pressures such as competing offers in the same city.
Limits of the model
The calculation above is a planning estimate, not an accounting truth. The learning-curve percentages should come from the process’s own data (handle time and quality by tenure month), not from assumptions. Some attrition is also healthy: a process with near-zero attrition may be holding on to low performers or offering no growth path. The goal is not zero, but a rate where the cost is understood and the early, avoidable exits are under control.
Written by the RP Strativa Team. Figures in the examples are illustrative.
Leave a Reply