RecruitingHR techAI screeningIndia

Applicant screening in two passes: filters, then a score

HireQwik November 13, 2026 10 min read

In 2021, Harvard Business School and Accenture published Hidden Workers: Untapped Talent. As HR Dive reported, 88 percent of the employers surveyed said qualified, high-skilled candidates were rejected outright because they did not exactly match the hiring criteria. For middle-skilled candidates, the figure was 94 percent.

Those aren’t candidates’ complaints. They’re employers describing their own filters. The mechanism is easy to recognise once you’ve run a high-volume role: a knockout question that was a little too strict, a keyword the résumé didn’t happen to use, a year-count that cut off a strong 18-month performer. Each filter seems reasonable on its own. Stacked, they quietly throw away the people you wanted.

This post is about applicant screening done in the other order: screen in first, by a score that reflects the job, and screen out only on the few conditions that are truly non-negotiable. The scoring half of the method is covered in depth in our guide to the candidate scoring model, the hub of this five-post series.

What is applicant screening?

Applicant screening is the first stage of hiring, where every application is checked against the role’s requirements to decide who goes on to an interview. Done well, it uses a small number of yes-or-no checks for true deal-breakers, then ranks everyone who passes by how closely their experience matches the job.

Search for the term and Google’s AI summary lists the usual parts: résumé review, knock-out questions, phone screens, assessments and background checks. That’s the full funnel. This post covers the first two, because they decide who ever reaches the rest.

Screening out vs screening in

The two approaches start from opposite questions.

Screening out asks: what disqualifies this person? It’s a chain of filters. Fail any one and you’re gone. It’s fast, it’s easy to automate, and it’s how most applicant tracking setups behave by default.

Screening in asks: who is strongest for this job? It’s a ranking. Nobody is removed for missing one thing; they’re placed lower. It’s slower to set up, because you need a scoring model, but it keeps strong-but-unusual applicants in view.

Screening outScreening in
Core questionWhat disqualifies?Who fits best?
OutputPass or failA ranked list
Main riskRejecting qualified peopleInviting too many if the cut-off is loose
Best forLegal and practical deal-breakersSkills and experience
Failure you won’t seeThe strong candidate filter 4 removedA weak model ranking the wrong things

Neither is enough alone. Pure screening out produces the 88 percent problem. Pure screening in ranks a candidate who can’t work the shift as highly as one who can, and you find out in the interview. The answer is both, in a specific order, with a hard limit on the first one.

Pass one: screen out on two or three real deal-breakers

The first pass removes only the applicants you could never hire. To qualify as a deal-breaker, a condition has to pass two tests: the hiring manager would refuse an otherwise excellent candidate who failed it, and an applicant can answer it with a clear yes or no.

Conditions that usually pass both tests:

  • Can work at the stated location, or from the stated region for remote roles.
  • Can work the stated shift or hours.
  • Has a licence or registration the law requires for the work.
  • Can join within the window the business needs.

Conditions that usually fail one of the tests: a degree, a minimum number of years, a named employer, “excellent communication”, a specific tool. Those are scored, not filtered. If you’re unsure which list a condition belongs on, sorting your job specification walks through it line by line.

Three rules for writing the deal-breakers themselves:

  1. Ask about the condition, not a proxy. “Can you work from our Pune office five days a week?” not “Are you based in Pune?” A candidate in Mumbai who will relocate passes the first and fails the second.
  2. Never reject on silence. A blank answer or an unreadable application is a reason for a person to look, not a rejection. Automating “no answer means no” is how good people disappear without anyone deciding to remove them.
  3. Cap it at three. Every extra filter multiplies the false rejections. If you have six, you have three deal-breakers and three preferences in disguise.

HireQwik takes these rules as ordinary sentences (“can work night shifts”), and it rejects nobody on its own: a candidate is turned down automatically only when HR turns on auto-decide for the job or adds deal-breaker rules to it. That default matters: filtering is a choice someone makes per job, with the rule in writing, not something the system does on its own. We went deeper on writing these in designing knockout questions that protect recruiter time.

Applicant screening questions: good and bad wording

The wording of a deal-breaker question decides whether it filters the right people. Most bad screening questions are bad in one of four ways: they ask about a proxy, they invite a hopeful yes, they have no clear right answer, or they ask something the law or basic fairness says you shouldn’t. Here are common ones, rewritten.

Weak wordingProblemBetter wording
“Are you based in Pune?”Asks about a proxy; excludes people who will move“Can you work from our Pune office five days a week from your joining date?”
“Are you comfortable with shifts?”Invites a hopeful yes“This role works rotating shifts, including two night weeks a month. Can you work that pattern?”
“What is your notice period?”No right or wrong answer as written“Can you join on or before 15 January?”
“Do you have excellent communication skills?”Everyone says yes; not a deal-breakerRemove. Score communication from the résumé and test it in the interview
“What is your age?” or “Are you married?”Not job-related; invites biasRemove
“Do you have a degree?”Usually a proxy, rarely a true deal-breakerRemove unless a law or licence requires it

Two more rules for the form itself. Put the deal-breaker questions first and keep them to three, so an applicant who can’t meet them finds out in a minute instead of after a 20-field form. And make every one of them mandatory with a plain yes or no, so the blank-answer problem from earlier mostly disappears.

If a question would need a paragraph to answer honestly, it isn’t a deal-breaker. It belongs in the interview.

Pass two: screen in by a job-matched score

Everyone who clears pass one gets scored against the job. The score should come from the criteria the hiring manager actually cares about, with weights, and each point should trace back to a line in the résumé. Because HireQwik quotes the résumé line behind each point it awards against the job’s requirements, the ranking can be checked rather than taken on trust.

The score then splits the list into three bands:

  • Invite: above the upper cut-off. These go to interview without another read.
  • Review: between the cut-offs. Someone reads each of these, with the evidence beside the number.
  • Decline or hold: below the lower cut-off.

How many applicants to screen in

The upper cut-off should come from capacity, not from a score you like the look of. Work it backwards from the hires you need. If the role needs 10 hires, and your last few drives turned about one in eight interviews into a hire, you need roughly 80 interviews. Add a margin for people who never book or don’t turn up; in our own data, 21 percent of booked interviews never start. So invite around 100, and set the upper line wherever the 100th score falls.

Then sanity-check the line by reading: five résumés just above it and five just below. If you can’t tell the two groups apart, the problem is the weights, not the line. If the top 100 still look thin, the problem is the applicant pool, and no cut-off fixes that; it’s a sourcing conversation, not a screening one.

The review band is where screening in earns its keep. The candidate with 18 months of quota-carrying sales and no degree lands here instead of in the rejected pile. So does the career-changer whose last title doesn’t match but whose described work does. A recruiter reading 80 résumés with highlighted evidence can make those calls well. A filter making them silently can’t.

What the two passes look like on a real-sized role

Here’s a hypothetical drive to show the shape: a customer support role, 1,000 applications, one recruiter, interviews to start within the week.

Day 1, morning. Applications arrive through a connected Google Sheet, or as a batch of uploaded PDF and Word files. The three deal-breakers (Pune office, rotating shifts, join within 30 days) run first. Say 260 fail one clearly and 40 left an answer blank. The 40 go to a person, not out.

Day 1, afternoon. The remaining 700 are scored against five weighted criteria. Upper cut-off set where the team’s interview capacity runs out; lower cut-off set where relevance clearly falls away.

Day 2. The recruiter reads the review band and the 40 blanks, with evidence in front of them. A few hours, not a week.

Day 2, evening. The invite band, plus whoever the recruiter pulled up from review, is approved together from the recruiter inbox. Each candidate is emailed a booking link to choose a time, with reminders as the deadline nears.

The numbers will be different for your role. The shape won’t: a short, strict first pass; a scored second pass; human reading only where the evidence is genuinely mixed.

Check what your filters removed

Screening out has a failure you can’t see unless you look for it: the person it removed never complains, because they never hear back. The only way to know your filters are right is to read a sample of what they rejected.

After each drive, pull 20 applications that failed pass one and 20 from the bottom band of pass two. Read them as if they’d just arrived. For each, ask one question: would the hiring manager have wanted to talk to this person? If more than two or three of the 40 are a yes, a filter or a weight is wrong. Find which one and fix it before the next drive. We described a version of this check for the interview stage in the No Go audit for catching false rejects.

Then tell the rejected applicants. A short, polite decline costs nothing to send in bulk and is the difference between a candidate who applies again and one who tells their friends you never replied, which is the problem we wrote about in the silent rejection.

Common applicant screening mistakes

Filtering on keywords. A résumé that says “handled customer escalations” and one that says “managed client complaints” describe the same work. A keyword filter passes one and rejects the other. Score meaning, not strings.

Letting the form do the filtering by accident. A required field the candidate couldn’t fill sensibly (“current CTC” for a fresher) becomes a silent filter. Review every required field as if it were a knockout, because it is.

One cut-off instead of two. A single line forces a pass-or-fail call on the candidates closest to it, which are exactly the ones that need a person.

Screening for the last hire. “We need someone like Priya” turns into filters that describe one person’s résumé. Describe the work Priya does instead, and score that.

From screened list to interview

Once both passes are done, the shortlist is the invite band plus whoever the recruiter pulled up from review. Moving them to interview should take one action, not a day of emails. Those who book sit a browser-based AI voice interview of 10-20 minutes, no download needed. It’s recorded and transcribed, so anyone on the hiring team can go back to the candidate’s exact words, and it closes with one of four recommendations: Strong Go, Go, On Hold or No Go. HR still decides.

If your pool of past applicants is large, the following post explains how to re-score past applicants before advertising.

Count your current hard filters for one live role today. If there are more than three, move the extras into the score. To watch both passes run on a real applicant list of yours, ask for a demo. To feel the interview from the candidate’s side, try it live.

Frequently asked questions

What is applicant screening?

Applicant screening is the first stage of hiring, where applications are checked against a role's requirements to decide who moves on to an interview. It usually combines a few yes-or-no checks, such as location or work authorisation, with a review or score of each résumé against the job.

What is the difference between screening in and screening out?

Screening out removes applicants who fail a requirement, so the question is what disqualifies someone. Screening in ranks applicants by how well they match the job, so the question is who is strongest. Screening out is fast but rejects good people when the filters are too many or too strict; screening in keeps them in the running.

How do you avoid rejecting qualified applicants during screening?

Keep hard filters to the two or three conditions you truly cannot hire without, write each as a yes-or-no question, and send anything ambiguous to a person instead of rejecting it. Then rank everyone who passes by a job-matched score, and check a sample of rejected applications after every drive.

See your own candidates screened

Book a 30-minute demo. Bring a live JD and we'll screen against it, then start with a pilot on your own candidates before committing to anything.