Your Screen Never Finds Out When It Was Wrong

Ask a recruiting team how well its screening works and you'll usually get a number about speed: time to first call, time to shortlist, maybe a pass-through rate. Quality of hire is the metric every TA leader says matters most, yet the screen, where most candidates actually get decided, almost never finds out how its decisions turned out.

Here's how it normally goes. A recruiter screens a picker on a Tuesday, the client starts them the following Monday, and five weeks later they stop turning up. If anyone records a reason, it goes into the client's HR system or becomes a payroll termination code. The recruiter who ran the screen never sees it, and the next Tuesday they ask the next applicant exactly the same questions.

This isn't rare. In a BambooHR survey, 31% of people said they had quit a job within the first six months [1], and in hourly, shift-based work it tends to be worse. Every one of those exits got a yes from somebody's screen. I've yet to meet a team that goes back and checks which of those yeses were wrong.

Where the answer ends up

Part of the problem is plumbing. The ATS owns the candidate until they're hired, and the HRIS or payroll system owns them after that. At a staffing agency the split is sharper still, because the reason someone left often lives with the client. So the question "did our screen see this coming?" needs data from two or three systems that don't talk to each other, and the person who could answer it is busy screening this week's applicants.

For agencies it's also a direct cost. A permanent placement that falls off inside a 60 or 90 day guarantee means a refund or a free replacement, which is a second search nobody pays you for. On temp work, you can't bill for a shift nobody works. Neither shows up in time-to-fill, which is part of why we think time-to-hire is the wrong clock.

The other part is the notes. Even when someone does pull the early leavers and open their screening records, what they find is "good energy, reliable, has transport." You can't learn anything from that. Nobody can say whether the recruiter asked about the commute, or asked and heard a vague answer, or just assumed.

The early exits are usually boring

When a team does manage to line things up, the reasons people leave in the first couple of months are rarely a surprise. The commute was longer than it sounded. The shift pattern didn't fit around childcare. The pay was a dollar under what they'd been told somewhere else. They had never done a job on their feet for ten hours before. None of this is hidden. It's the kind of thing a candidate will tell you if someone asks properly and follows up.

"Has transport" and "takes two buses to get to a 5am start" are both true of the same person. Only one of them tells you anything.

So the exercise I'd suggest is small. Pick one high-volume role. Pull everyone you hired into it over the last six months and mark who left within 60 days. Then go through the screening answers one question at a time and compare the two groups. With these sample sizes you're not doing statistics. You're looking for gaps big enough to see by eye.

Some questions will show no difference at all, and those can probably go, which makes the screen shorter for everyone. One or two will separate the groups clearly, and those deserve a better follow-up. This only works if every candidate was asked the same questions in roughly the same way. If each recruiter runs the call their own way, and most do, there's nothing to compare.

How Asendia AI handles this

We built Asendia as a voice-first AI recruiter. It phones every applicant, usually within hours of them applying and at any hour, 24/7, which counts for a lot when your candidates work nights. Every call follows the same structure for the role, and when an answer is vague it asks again. "Yes, I can get there" gets a "how would you get to a 5am start?"

Each answer goes into the ATS you already use as its own field on the candidate's record, next to a short summary, not buried in a paragraph of notes. That's what makes the exercise above practical. When you pull your 60-day leavers, the commute answer, the shift answer and the pay expectation are sitting there in the same format for everyone, so comparing them is an afternoon with a spreadsheet, not a project.

You can point the same kind of call at people after they start. A check-in during someone's second week, asking how the shifts and the journey are going, tends to catch the problem while it can still be fixed with a shift swap or a different site, not after a no-show. Most of the agencies using Asendia came to it for volume, to screen hundreds of applicants a week without adding recruiters. The slower payoff is a screen that gets a little better every quarter, because it finally hears when it was wrong.

Final Word

Speed metrics tell you whether your screen is fast. Early attrition is the only thing that tells you whether it's right, and at the moment that data sits in somebody else's system where nobody in recruiting looks at it.

Try the smallest version this month. One role, six months of hires, a column for who left inside 60 days. Then read the screening notes for the people who left. If you can't tell from those notes why any of them would go, you know what the screen needs to start asking.

Ready to transform your hiring strategy? Schedule a Demo with our founders today!: https://asendia.ai/talk-to-founders