When a role gets four hundred applications, the instinct is that screening is a volume problem. There are too many résumés and not enough people, so you read faster, skim harder, or wait until someone has a free afternoon.
The volume is real. LinkedIn’s research in India found that applicants per open role have more than doubled since early 2022. But volume is rarely what makes screening slow. What makes it slow is reading each résumé without having decided what you’re looking for. Every one turns into a small judgement call, and judgement calls are slow, tiring and inconsistent by the end of the afternoon.
Fast screening is mostly decided before you open the first résumé.
Decide what “yes” looks like, with the hiring manager
Most job descriptions list a dozen or more requirements and rank none of them. The recruiter is then left to guess which ones matter, and the hiring manager rejects the shortlist a week later because the recruiter guessed differently.
So start with a short conversation. Split the JD into three groups: the things a candidate genuinely can’t do the job without, the things that would put one good candidate ahead of another, and everything else. That last group, “premier institute”, “excellent communication skills”, a list of certifications, is usually larger than people expect. Either it can’t be judged from a résumé or it doesn’t predict the job. Agree to ignore it.
The must-haves should be few: three to five. If the hiring manager insists on eight, ask which three they’d give up for an otherwise excellent candidate. Those three weren’t must-haves.
Then test the agreement before you rely on it. Screen the first ten or so résumés, send them to the hiring manager with your call on each, and talk only about the ones where you disagree. A disagreement usually means there’s a requirement nobody wrote down (“I’d never hire someone without on-call experience”), or one that isn’t really a requirement after all. Twenty minutes here saves the week you’d otherwise lose to a rejected shortlist.
Let candidates do the first filter
Almost every job board, and any decent application form, lets you ask a few required questions when someone applies. Use them for the must-haves that a plain answer can settle: years of relevant experience, whether they’ve done the core work, whether they can work from your location. Anyone who clearly doesn’t meet a must-have is filtered out before a recruiter spends a second on them.
Keep it to a handful of questions, because every extra one costs you applicants. Only filter on must-haves, never on college, age, gender or anything that stands in for them. And check what gets filtered out in the first few days, because a badly worded question can quietly reject exactly the people you wanted.
Read in two passes
Screening is slow when every résumé gets the same careful read. Most don’t deserve it, in either direction.
The first pass asks one question: does this person have the must-haves? It should take seconds, not minutes. You aren’t ranking anyone yet and you aren’t reading the summary paragraph. Recruiters already skim: Ladders’ eye-tracking research found an average initial look of about 7 seconds per résumé. Skimming is fine when you know what you’re skimming for. When the evidence might be there but isn’t stated clearly, put the résumé in a “not sure” pile rather than rejecting it.
The second pass is where the time goes, and it’s spent on far fewer people. Now you compare: how strong is the evidence, and how many of the nice-to-haves does each person bring? Write one line on why each shortlisted candidate made it. That line is what you send the hiring manager, and what you’ll want in three weeks when someone asks why candidate 214 wasn’t called.
Specifics beat polish
Résumés look more alike than they used to. In a recent report on Indian hiring, more than half of the recruiters who said hiring had become harder blamed an increase in AI-generated applications. Trying to spot which ones a tool wrote is a losing game, and it isn’t really the point. Plenty of strong candidates use writing tools, and plenty of weak ones always had a friend write their CV.
What polish can’t fake is detail. “Improved system performance” could be anyone. “Cut checkout latency by moving reconciliation to a nightly batch” is a person who did something. Look for named systems, numbers the candidate would know, trade-offs they made, things that went wrong. A generic résumé goes to the “not sure” pile, and a short screening call or a small, relevant task settles it far better than staring at the wording.
Gaps, colleges and other false alarms
Speed pushes people towards shortcuts, and most shortcuts are biases with a respectable name. College tier. A gap in 2023. Three short stints at startups. English that reads as a second language. None of these tells you whether someone can do the job, and screening on them shrinks the pool for no gain. In LinkedIn research reported by UNLEASH, 73% of US respondents said they had taken a career break at some point. A gap is a fact to ask about, if it matters, not a verdict.
The real warning signs are different: dates or titles that contradict each other, skills listed with no work anywhere that used them, answers to the application questions that don’t match the résumé. Even those are questions for a call before they’re reasons to reject. Sometimes a mismatch is a typo.
The same goes for software. Tools that learn from past hiring decisions can learn their bias too. Amazon scrapped an experimental screening tool after it penalised résumés containing the word “women’s”, and a 2024 University of Washington study found that the language models it tested favoured white-associated names 85% of the time when ranking otherwise identical résumés. Whatever does the first read, a person should own every decision, and someone should regularly look at who is being screened out.
Notice periods deserve a word of their own in India. A strong candidate on 90 days may well be worth the wait, so record the notice period and plan around it rather than filtering on it. Then keep them warm through to joining.
What breaks when it’s all in folders and spreadsheets
Here’s how screening usually runs without a system. Résumés are downloaded from two job boards into a folder, a few more arrive by email from referrals, and a spreadsheet tracks who’s been seen, colour-coded in a way only its owner understands. The same candidate applies through two sources and gets two different answers. The recruiter is out for a week, and nobody else can tell which 60 of the 400 were read. The hiring manager’s must-haves live in a WhatsApp message from the kickoff call.
None of this is anyone’s fault. It’s just that the decisions that make screening fast (what you’re looking for, why each person was or wasn’t shortlisted) end up scattered in places nobody can search. Each new role starts from scratch, and the reasons for each decision are gone by the time anyone needs them.
Good screening needs the criteria written down once, applied the same way to every résumé, and a reason recorded against every candidate. That’s exactly the kind of repetitive, consistent work that a system does better than a person on their fourth hundred résumés.
How Arexis helps
Arexis reads the job description and splits it into must-haves and nice-to-haves, then scores every résumé against the role, with the evidence for each score shown alongside it. Résumés are anonymised before they’re scored, so the first read is about the work, not the name or the college. Your team reviews the ranked list and makes every call, with the reasoning kept on each candidate. See how screening and scoring work.
Found this useful? Pass it on.