Zapier Freed 84 Recruiter Hours With Voice AI Screening

Correction, 3 September 2026: an earlier version credited Deloitte with a “23% faster time-to-hire” figure and Zapier with a “30% recruiter time” figure. The offer, start and retention gains come from the University of Chicago Booth field experiment; Zapier’s result, as reported by Benefit News, is 84 recruiter hours returned in the pilot. A “31% have no AI plans” figure attributed to SHRM was replaced with what SHRM published. SHRM’s 2026 State of AI in HR analysis found that roughly a third of organisations have no AI adoption at all. This post is for them.

Not a forecast. Not a prediction. Not an argument about the future of work. Just third-party receipts from companies that have run voice AI screening at scale and measured what happened.

If your team is still evaluating whether voice AI screening works before committing to it, these numbers are what you need to see.

Zapier: 84 Recruiter Hours Back From One Pilot

Zapier partnered with Ezra to implement AI voice screening across their high-volume hiring funnel. The outcome, reported by Benefit News in March 2026: the pilot returned 84 hours of recruiter capacity, and Zapier projects 5 to 6 hours back per recruiter per week if it switches fully, with Ezra’s scores tracking its own recruiters’ assessments closely.

That’s not a projection. That’s the measured difference between their pre-AI workflow and their post-AI workflow, tracked across a real hiring cycle with real candidates.

The mechanism is straightforward: AI screening handles the first-round conversation, classifies candidates, and delivers a prioritised shortlist to recruiters. Recruiters spend their time on the shortlist — candidates worth advancing — rather than on the full application pool.

Thirty percent of a recruiter’s time is approximately 12 hours per week. At a 45-week working year, that’s 540 hours per recruiter per year recovered. For a 5-person TA team, that’s 2,700 hours — roughly 1.5 full-time equivalent headcount — redirected from screening calls to work that actually requires human judgment.

Chicago Booth: 12% More Offers, 18% More Starts, 17% Higher 30-Day Retention

A University of Chicago Booth field experiment by Brian Jabarian and Luca Henkel randomly assigned about 70,000 applicants for entry-level customer service roles to a human recruiter or an AI voice interviewer. Three numbers stood out:

18% more job starts. The gain comes from eliminating scheduling friction (AI screens are available immediately, not scheduled 5–7 business days out), eliminating recruiter availability constraints, and delivering a classified shortlist rather than raw interview notes.

12% more offers extended. This is the counterintuitive finding. More candidates completing the first-round screen — because the AI is available evenings and weekends, not just business hours — means a larger qualified pool reaching the offer stage. The process captures candidates it would have previously lost to scheduling friction.

17% higher 30-day retention. This one requires a moment to understand. The hypothesis: AI screening that focuses on communication fit, role understanding, and values alignment — rather than just keyword matching — produces hires who have a more accurate picture of the role before joining. Better pre-joining understanding correlates with better early retention.

Together, these three metrics describe a complete pipeline improvement: faster at the top, more efficient in the middle, better outcomes at the bottom.

HireQwik Pilot Data: 1,099 Interviews, 14 Campaigns

Third-party case studies establish that voice AI screening works. HireQwik’s own pilot data establishes the specific operational parameters for India’s campus hiring context.

1,099 interviews conducted across 14 campaigns. These are live HireQwik deployments at Indian companies running campus hiring drives in the April–June and August–October peak seasons.

18 hours → 1–2 hours review time. The consistent finding across pilots: HR teams that previously spent 18 hours reviewing first-round screen summaries, coordinating recruiter notes, and building shortlists completed the same step in 1–2 hours using HireQwik’s classified, transcript-backed shortlist.

Up to 1,300 interview slots a day. HireQwik’s capacity-aware slot booking spreads a drive across the day without queue delays. A 3,000-candidate drive can be screened in 2 days rather than spread across 6–8 weeks of recruiter scheduling.

60%+ auto-rejection rate. More than 60% of candidates self-select out or are classified No Go in the first screen — before any recruiter time is invested. This is the filtering function that makes the downstream economics work.

What the Numbers Say Together

The pattern across all three data sources is consistent: voice AI screening compresses time at the top of the funnel, improves the quality of candidates reaching the offer stage, and produces better downstream outcomes without adding recruiter headcount.

This is not a pilot-phase technology. Zapier is a global company. The Booth numbers come from a 70,000-applicant randomised experiment. HireQwik’s pilot data comes from Indian companies running real campus drives under real peak-season constraints.

For the Third Who Haven’t Started Yet

SHRM’s finding that roughly a third of organisations have no AI adoption cuts both ways. If you’re in that third, you’re not behind yet — but the window is narrowing. The companies in your talent market who have deployed AI screening are compressing their hiring timelines, recovering recruiter capacity, and extending offers faster. That’s a direct competitive disadvantage in any market where the best candidates have multiple options.

The receipts are in. The question is whether your team collects them in FY27 or FY26.

If you want to see HireQwik’s pilot data in detail — the specific metrics from the 14 campaigns, broken down by drive size and industry — connect with the team and we’ll send the full breakdown.