The irrelevant-applicant problem: why one opening pulls thousands of resumes that don't fit

Post a campus operations role on Naukri today and you’ll have 2,500 applications by Thursday — the majority from candidates who’ve never done operations work and may not want to.

This isn’t a 2026 anomaly. It’s a structural feature of how Indian fresher hiring works, and it consumes most of a recruiter’s productive day before a single qualified candidate picks up the phone. The scale has compounded steadily since India hired 1.2 million+ campus freshers in the 2024–25 cycle (NASSCOM), and job aggregators have made one-click applying frictionless enough that a student submitting 200 applications in a single afternoon is unremarkable.

Why the mismatch isn’t the candidate’s fault

When 73% of Indian employers plan to hire freshers in 2026, and each posting goes live on four or five aggregators simultaneously, every campus student faces the same rational calculation: apply broadly and let the recruiter filter. The system rewarded this behavior by keeping application costs near zero. The recruiter side absorbs the full downstream cost.

The mismatch compounds in ways that aggregate numbers hide. A tier-1 NIT graduate applying to a ₹30K CTC data-entry role inflates the applicant count while pushing the genuinely matched mid-tier candidate further down the stack. A civil engineering student applies to a software testing role because the JD says “any engineering branch.” That JD was written to maximise sourcing breadth — it worked — and the hiring team now owns the consequence.

The sourcing optimisation trap

The first instinct is to screen faster. The actual problem is that most JD language is optimised for search indexing, not qualification filtering. “Strong communication skills,” “willingness to learn,” and “analytical mindset” are phrases every applicant technically possesses. They don’t filter. They invite.

When you write a JD for discoverability and then complain about irrelevant applicants, you’ve built a funnel whose mouth is four lanes wide and whose neck is a straw. The aggregator delivers exactly what you asked for: maximum reach. You then absorb the cost of sorting what maximum reach produces — a cost that scales linearly with the size of the opening and the strength of your employer brand.

The contrarian take here: the irrelevant-applicant problem is largely self-inflicted. Most TA teams write JDs for sourcing breadth and design screening flows as if they’ll receive a curated set of matched candidates. These two assumptions cannot coexist at the 2,500-candidate scale.

Why volume compounds in India specifically

The challenge is specific to Indian campus hiring in a way global benchmarks miss. A tier-3 institution in Bhopal produces graduates with genuine capability but low brand recognition, so those students apply to everything, everywhere, in volume. A tier-1 IIT produces graduates who apply broadly as a hedge and decline the majority of offers. Both cohorts inflate your applicant count in ways that look identical on a volume dashboard but require completely different handling at the screen.

The 1:30 HR-to-candidate ratio that campus hiring teams in Indian IT services routinely hit is a symptom of this heterogeneity — not just of volume. Recruiter bandwidth is finite. Spreading it across 3,000 applications, most of which are misdirected, means the matched candidates buried in the middle of that pile get less attention than they deserve.

What shifts the ratio

The fix isn’t better resume parsing. Resume scores cluster regardless of how sophisticated the matching algorithm is, because resumes at the freshers tier are fundamentally low-information documents — everyone has roughly the same signals. The actual fix is moving the first filter earlier and making it conversational and role-specific.

A structured screening interaction derived from the specific JD’s requirements asks candidates to demonstrate something rather than describe something. A question built from what the role actually needs separates genuine mismatch from genuine fit faster than any keyword scan. Someone applying to a data-analyst role from an unrelated background who can’t answer a basic question about the data they’d be working with self-selects out quickly. Someone who’s been quietly building the relevant skills but buried them three lines deep in their resume suddenly stands out.

In enterprise pilot campaigns running at the 2,500–3,000 candidate scale, phase-0 knockout questions — designed to fire in the first 60–90 seconds of a structured screening call — surfaced this mismatch signal within the first two minutes of each conversation. The 15–20 minute structured screening conversation that followed was reserved for candidates who’d cleared the initial fit check.

For a practical breakdown of how the first-filter question set is structured for exactly this purpose, the knockout-questions post covers phase-0 design in detail.

The cost of not fixing this

Across our pilot of 1,099 interviews completed, the reduction in HR time per candidate versus manual phone screens reached 89%. But that number is only meaningful if the inputs to the filter are calibrated. Running a fast screen on 3,000 misdirected applicants produces fast noise.

The SHRM 2025 AI-in-HR survey found that 88% of HR leaders cite AI screening as a compliance risk. Most of that audit exposure concentrates at the rejection stage. A structured, role-specific first filter produces a defensible audit trail. A high-volume keyword sweep on misdirected applicants produces exposure without corresponding benefit.

The irrelevant-applicant problem is expensive and demoralising for the TA team — but it isn’t inevitable. The answer isn’t to slow down applications; that option went away when aggregators commoditised one-click applying. The answer is to front-load a signal-extraction step that costs candidates two minutes and costs recruiters nothing in aggregate.

Fix the mouth of the funnel. The numbers downstream follow.