How to screen two hundred applicants for one role

Start with the arithmetic, not the advice

Two hundred applicants is not a large number in the abstract. It becomes a large number the moment somebody has to speak to each of them. A first-round screening call that recruiters describe as 20 minutes is rarely 20 minutes: there is the attempt that goes to voicemail, the reschedule, the note written afterwards. But take the optimistic figure and hold it.

At 20 minutes each, 500 screenings is 167 hours, or about 21 working days of somebody doing nothing else. On a recruiter costing an employer RM 6,000 a month, which is roughly RM 34 an hour once statutory contributions are counted, that is RM 11.36 of payroll per screening and about RM 5,680 for the batch.

Those figures are ours and they are arguable. The point is not the total. The point is that the first round has a fixed cost per candidate, and every plan for handling volume is really a plan for changing one of three numbers: how many candidates reach the first round, how long each one takes, or who does it.

Why the usual fixes disappoint

Keyword filtering the pile. Screening on CV keywords is the fastest way to cut two hundred to forty, and it is the reason good candidates are invisible. A CV reflects how well somebody writes a CV. It answers almost none of the questions a first round exists to answer, and it is silent on all four of the ones that most often end a process: what they expect to be paid, when they can start, whether they are still available, and whether they have actually done the thing.

One-way video. Recording answers to fixed prompts moves the hour off the recruiter, but it moves it onto the candidate, and it moves it in the most expensive possible form. It asks for a camera, a quiet room, decent upload speed and a willingness to perform. Drop-off is the predictable result, and it does not fall evenly: the people who abandon it are disproportionately the ones without a private room to sit in.

Longer assessments. An assessment that takes an hour is a filter on free time before it is a filter on ability. It is defensible for a final shortlist. As a first gate on two hundred people it selects for the unemployed and the underemployed.

Hiring another recruiter. This genuinely works, and for a sustained increase in volume it is often the honest answer. It is worth saying plainly rather than pretending software is always the cheaper choice. What it does not do is flex: the pile arrives in a fortnight and a hire takes a quarter.

What actually reduces the pile

Ask the disqualifying questions first, and ask everybody. Most first rounds are structured as an interview with the eliminating questions at the end. Reverse it. Salary expectation, notice period and current availability decide more outcomes than anything else in the conversation, they take two minutes, and asking them of every applicant rather than of a shortlist means the shortlist is drawn from answers instead of from CVs.

Write down what a good answer looks like before you start. Not a score out of ten, which people apply differently on a Friday. The specific thing you need to hear. Doing this once, in advance, is what makes it possible for a first round to be run by somebody other than the hiring manager, or by software, without the standard drifting between candidate one and candidate two hundred.

Make the first round asynchronous but conversational. The scheduling is usually a larger cost than the conversation. A first round that a candidate can open when they have ten minutes, and that can ask a follow-up when an answer is vague, removes the calendar without removing the ability to probe. That combination is the one thing a form cannot do and a phone call does expensively.

Keep the transcript. Whatever runs the first round, the output that matters is what the candidate actually said, not a rating. A score with no transcript under it cannot be checked, cannot be appealed, and cannot be handed to a hiring manager who wants to know why.

The order to do it in

If you are in the middle of a pile right now, the sequence that recovers the most time for the least change is: write the three or four questions whose answers would let you say no with confidence, ask them of everyone before anybody books a call, and reserve live conversations for the people whose answers survived. That alone moves most of the cost from 500 conversations to a fraction of them.

The structural version of the same idea is to stop treating the first round as a small interview and start treating it as intake. Intake is repeatable, and repeatable work can be handed over.

The questions themselves are the part most worth getting right, and they are covered separately in the questions a first round actually has to answer. If you are considering handing the first round to software, the fairness question is the one to settle first, and that is in is an AI screening interview fair to the candidate.