Ask a hiring manager what a screening call costs and the answer is usually "nothing, the recruiter is already on payroll". That is the same logic that makes meetings free. It is worth doing the arithmetic once, properly, because the number is larger than people expect and it is the number every automation decision should be measured against.
The arithmetic
Take a recruiter on ₹6,00,000 a year. Roughly 2,000 working hours, so about ₹300 an hour in salary alone — before the laptop, the ATS licence, the desk, or the manager's time.
Now a single screening call:
- Dial, wait, no answer — 1 minute. This happens on more than half of first attempts.
- Second attempt later in the day — another 1 minute.
- Connected call — 6 minutes of actual conversation.
- Writing notes into the ATS — 3 minutes, if they do it immediately.
- Context switching back to whatever they were doing — call it 2 minutes.
That is 13 minutes of recruiter time per candidate actually reached, or about ₹65. For 400 applicants, most of whom are not reachable on the first try, you are looking at roughly 85 recruiter-hours and ₹25,000 of salaried time — to produce a shortlist of perhaps 40 people.
The part that does not show up in the number
The cash cost is the smaller half. The larger half is what those 85 hours were taken from. Recruiters do not have a queue of idle time; screening calls displace sourcing, candidate experience, and closing offers. The most expensive thing about screening is that your most experienced person is doing the most repetitive part of the job.
The second hidden cost is inconsistency. By call 200, the questions have drifted, the standard has moved, and the notes on candidate 12 and candidate 380 are not comparable. That is not a discipline failure — it is what happens to any human doing the same conversation four hundred times.
The comparison
The same 400 calls placed by a voice agent at ₹4.5 per two-way minute, averaging 5 minutes each, cost about ₹9,000 — and the no-answers cost nothing, because an unanswered call bills no minutes. Every one produces a transcript, a summary and a score against the same rubric.
The recruiter still does the work that needs judgement: reading the 40 that scored well, making the calls that need a human, closing the two that matter. What is gone is the 340 calls that were only ever going to end in "thanks, we'll be in touch".
The honest caveat
If you screen fifteen people a month, none of this applies and you should not automate anything. The arithmetic only turns at volume, because the fixed cost of writing a good prompt and a good rating rubric is real — call it half a day of work. Below roughly a hundred calls a month, that half-day never pays back.
Above it, the question stops being whether to automate and becomes which conversations still deserve a person. See the per-minute rates, or run the numbers for your own volume with the calculator on the home page.