AI Screening Cuts Hiring Time 43% — But Here's the Real Catch

The ROI case for AI screening has finally moved from vendor promises to independent data. A wave of research published between 2024 and 2026 — from Aptitude Research, Lighthouse Research & Advisory, SHRM, and WorkTech Academy — now gives us something far more useful than a case study: actual signal on where AI screening works, where the gains hold up, and where they quietly disappear. The headline number is 43%. That's the reduction in time-to-shortlist that Aptitude Research found across 350 enterprise hiring programmes in 2025. But the full picture is more nuanced — and for HR leaders making deployment decisions, the nuance matters more than the headline. In this episode, we break down where the genuine efficiency gains come from (early-funnel CV triage and first-round screening), why quality improvements at 90 days don't always persist to 12 months, and the two patterns that consistently erode AI screening ROI in enterprise deployments: deploying with default rubrics and ignoring candidate dropout rates. We also look at why modality matters — voice and conversational AI tools see 22–28% lower dropout than form-based or pre-recorded video alternatives, and how OVI's audio-only screening agent Milo is built around exactly that insight. If you're evaluating AI screening tools or trying to understand why your current deployment isn't delivering the ROI you expected, this episode is your starting point.