The Bottleneck Moves Downstream

Recruiting teams that start screening applicants with AI tend to run into the same surprise. Screening gets much faster, and hiring barely does.

It shouldn't be surprising. Anyone who has worked in a factory knows that when you speed up one step in a process, the slowness moves to the next step. For decades the slow step in hiring was at the top: too many applicants and too few recruiter hours to get through them. AI has started to take that one away, and the next one in line is the hiring manager's calendar.

Imagine a team that has just started screening with AI. Where it used to take two weeks to produce a shortlist, it now takes a couple of days. On Monday morning the recruiter opens a queue of a few dozen qualified candidates with conversation summaries attached, instead of a few hundred unread applications. Initial screening that used to eat three weeks is done by day three, and the team is ready to present candidates.

Then the hiring manager asks if they can find time next week.

That week is the new bottleneck. The screening software, the volume of applications, and the quality of the resumes all matter less now than the seven days it takes a busy manager to clear a slot between meetings. A process that used to be measured in weeks now spends most of its time waiting on one person's calendar.

Waiting a week sounds harmless. What makes it expensive is that candidates don't wait with you. Think about someone who applied on a Wednesday, had an AI screening call that evening, and was told on Thursday that they'd be interviewed. That person is about as interested as a candidate gets. They've put in real time, had a real conversation, and have a clear idea of what should happen next. On Friday they're expecting to hear from you.

If the manager's first open slot is the following Thursday, eight days later, a lot can happen in between. The candidate may interview somewhere else, maybe twice, and get follow-up calls from other companies. Each of those pulls a little of their attention away from your role. By the time your manager is free, you may be the company they're least excited about, and a week that felt routine on your side was long enough for someone else to make an offer.

Seen this way, the process runs at two speeds. Intake, first screening, and qualification have had years of automation, and that was a sensible place to start. But the stages after the shortlist are as manual and as tied to calendars as they were in 2015: scheduling the manager, preparing for interviews, making the decision, giving feedback. A fast front half feeds a human-paced back half, so the whole thing gets you to the next bottleneck sooner and not much further.

This is easy to miss because the number most teams watch is time to first screen, and that one gets better. The slow stretch, from first interview to offer, depends on how soon the manager gives feedback, how long several stakeholders take to agree, and internal alignment. AI hasn't touched any of that. So a team can look like it's winning on the number it measures while candidates wait as long as they always did.

What can you do about a constraint that is a person's calendar? You can't add hours to the manager's week. You can change how much of each interview goes to work that should have happened earlier. A lot of first hiring manager interviews are really discovery calls, where the first fifteen minutes go to finding out whether the candidate is qualified at all. If that's already known before the manager walks in, the interview can be shorter and go straight to judging the person. Feedback is easier to give, and decisions come sooner when you're comparing three well-documented candidates instead of deciding whether a PDF deserves a conversation.

That's where I think most of the value in AI screening is, and it's what we try to do with Asendia. We make software that phones applicants and interviews them. What reaches the manager is a shortlist of people who have already been talked to, with their answers on the key role criteria written down and quotes from their own screening conversation attached. The manager's first interview can then be an actual evaluation, which is where their time belongs.

For agencies it changes what they hand a client. Instead of sending a pile of candidates and leaving the client's team to sort them, they can send a pre-screened shortlist where the client's first conversation with each person is already an evaluation. The recruiter isn't working any faster, but part of the client's bottleneck has been dealt with before the shortlist arrives, so placements should move more quickly. Intake quality matters downstream for the same reason, and the post on AI-generated applications flooding your ATS looks at how problems at the top of the funnel spread into every stage after it.

Automating the recruiter's part of hiring was necessary, but it only gets you halfway. The manager's calendar isn't going to open up on its own. What you control is what you put in front of them, and a manager who is choosing between people they already know a lot about moves faster than one who is still finding out who they are. If AI has made the front of your process quick and the back hasn't changed, the back is the half to work on next.

Ready to transform your hiring strategy? Schedule a Demo with our founders today!