Advice and guides / For employers
How to screen four hundred CVs without reading four hundred CVs
The cost of a bad screening process is not the hours. It is the good candidate you rejected at step one and the weak one you interviewed at step four.
Four hundred applications for one role is now ordinary, and it will be five hundred for anything remote. Nobody reads four hundred CVs properly. What actually happens is that somebody reads the first sixty carefully, the next hundred quickly, and the rest not at all, which means the outcome depends heavily on when a candidate applied.
The fix is an order of operations where each step costs more than the one before it, and each step removes people for a reason you could defend out loud.
Step one: the hard constraints, and only the hard constraints
Location, notice period, salary expectation, work authorisation. These are cheap to check and they are genuinely binary. A candidate with a ninety day notice cannot fill a role starting in four weeks, and no amount of reading their CV changes that.
The discipline here is to keep the list short. Every extra constraint at step one removes people for something that might have been negotiable. Degree requirements and years of experience do not belong in this step.
Step two: evidence of the thing, not the word for the thing
Keyword filters fail in both directions. They pass the candidate who listed the tool and never used it, and they fail the candidate who described the work without naming the tool. If you must filter on keywords, filter generously and accept the noise.
Step three: make them do something small
This is where high volume screening either works or does not. A short structured task, the same for everybody, marked the same way. Half an hour is plenty. It costs the candidate a real thirty minutes, which is enough to drop people who are mass applying without reading, and it produces a comparable score across four hundred people instead of four hundred subjective impressions.
The single most useful thing you can put in it is two questions drawn from the candidate's own CV. A generic paper can be answered generically. A question about the order someone did their own month end close in, and the thing that most often went wrong, cannot.
Step four: sort, then read
Now you have a list you can order. Sort by whether the CV stood up to its own questions first, and by score second. Read the top fifteen properly. That is a morning, not a week.
Keep the rest. A candidate who was not right for this role at this band is frequently right for the next one, and a pool you can filter is worth more than a folder of CVs nobody will open again.
What to avoid
- Rejecting on gaps. Career gaps in India correlate with caregiving, illness and one bad employer, not with capability.
- College filters. They shrink your pool hard and predict very little outside campus hiring.
- Long take home tasks. Anything over an hour drops good candidates who already have a job.
- Different questions for different people. If the screen is not the same, the scores are not comparable and you have not sorted anything.
This is the process we run for every mandate. What you get back is described on what you receive, and the assessment itself is published section by section on how it works.