How We Screened 3,000 Candidates in 2 Hours

Campus hiring at scale is brutal. When your HR team faces 3,000 applicants for a single campus drive, the screening phase alone can consume 18+ hours spread across 4 working days. We built HireQwik to solve exactly this — and the results have been validated by enterprise HR professionals managing real campaigns.

The Problem: Manual Screening Doesn’t Scale

Manual screening does not scale because its cost rises with every applicant while a recruiter’s day does not. A 3,000-candidate drive takes around 18 hours of profile review spread across 6 to 10 working days, occupies two full-time HR team members, and loses roughly half the pool as soon as video submissions are required.

Here’s what a typical campus hiring campaign looks like without AI:

• 3,000 applicants per campus drive
• 18 hours of manual profile screening (75-90 minutes per 200 profiles)
• 6-10 working days for the entire screening phase
• 2 HR team members working full-time on screening
• 50% drop-off when video submissions are required

The math is simple but painful: at Rs.85-150 per manual phone screen, screening 3,000 candidates costs Rs.2.5-4.5 lakhs per campaign. Multiply by 9 campaigns per year, and you’re looking at Rs.22-40 lakhs annually — just for the screening phase.

What We Did: AI Voice Conversations at Scale

HireQwik replaces the manual phone screen with a 15-20 minute AI voice conversation. Here’s how it works:

1. Upload your JD and candidate list — the AI agent configures automatically based on your job requirements
2. Candidates receive an invite link — they join a browser-based meeting room (no app download needed)
3. The AI conducts a structured screening conversation — assessing communication skills, role relevance, and basic domain knowledge
4. Automatic classification — every candidate is sorted into Strong Go / Go / On Hold / No Go tiers

The key insight from HR professionals: “I want to check communication skills first. Then I look at role relevance.” Our AI follows this exact philosophy — communication-first filtering, not just resume keyword matching.

The Results: Validated by HR

After piloting with enterprise HR teams managing 2,500-3,000 candidate campaigns, here’s what we measured:

The quote that stuck with us: “Instead of it being a 4-day thing, it comes down to 4 hours. So it is like 10x faster.”

Why Voice Beats Video

One of the most important learnings: top candidates refuse to record video submissions. As one HR lead told us: “Top candidates won’t try to record themselves. But if it’s a meeting, 15-20 minutes, they will just interact.”

By replacing video submissions with voice conversations:

• Applicant pool doubles — from ~3,000 to ~6,000
• No scripting possible — real-time conversational probing defeats ChatGPT-prepared responses
• Zero candidate prep needed — just show up and talk

The Economics

At AI-screening unit costs:

• 27,000 screens/year (3,000 candidates x 9 campaigns) at AI-screening rates
• Manual equivalent would cost Rs.22-40 lakhs
• Net savings: 50-70% cost reduction + 288 hours of HR time reclaimed

The real ROI isn’t just cost savings — it’s that your HR team can focus on domain interviews, offer decisions, and strategic hiring work instead of spending 4 days on initial screening.

What this case study doesn’t show you

A single evening of throughput is the easiest part of this story to tell and the least useful part to copy. Three things we would want a TA team to know before reading the numbers above as a promise.

The extreme bands were clean. The middle was not. Strong Go and No Go held up under recruiter review. On Hold did not sort itself — those candidates passed the screen but showed mixed signals, and someone still has to read the transcript to decide. Budget review time for the middle band, not for the whole pool.

Accent and code-switching are still edge cases. Pure-English conversations score well. Heavy regional code-switching and Hinglish produce cases we score conservatively on purpose, because rejecting one good candidate is not recoverable the way selecting two wrong ones is. Any vendor claiming that problem is solved, including us, is overclaiming.

Trust is earned per role, not per platform. A rubric calibrated for one role will reject strong candidates in another, and no HR team should approve an AI’s rejects in production before spot-checking around fifty audio-recorded conversations themselves. The 10x number above is what the process looks like after that calibration work, not instead of it.

Key Takeaways

1. Communication-first filtering works — HR professionals validated that assessing communication before resume relevance catches the right candidates
2. 60%+ auto-rejection is the magic number — if the AI can confidently reject clear mismatches, HR only reviews qualified profiles
3. Remove video submission friction — the best candidates won’t jump through hoops, but they’ll happily have a conversation

Want to see your numbers? Try our ROI Calculator to calculate your specific savings based on your hiring volume and team size.