AI Sourcing and the Passive Candidates Nobody Calls

Give a recruiting team a tool that finds five times as many good candidates for every role, and ask how many more people they'll hire.

For most teams, I suspect not many. The reason why says something about where the hard part of recruiting actually is.

That tool more or less exists now. AI sourcing has been one of the fastest-adopted kinds of recruiting software over the past two years. LinkedIn's AI recommendations, Gem, Beamery, Findem, SeekOut: the tools for finding passive candidates who fit a role profile have become good and fast, and they cost less than paying sourcers to do the same work by hand. A search that used to take a senior researcher three days can be set up in 30 minutes. The software reads LinkedIn profiles, GitHub commits, portfolio sites and professional databases at once, ranks what it finds, and drafts outreach that doesn't read like a template.

What it can't do is talk to anyone.

Response rates for passive outreach are usually put somewhere around 10 to 15 percent. So a campaign aimed at 1,000 people might get 100 to 150 interested replies. That sounds manageable until you remember who has to answer them. It's the same team that has 200 inbound applicants to screen, three roles in final interviews and two offer letters to write. An interested reply isn't pipeline yet. It sits in the outreach tool until a recruiter finds fifteen minutes for a call.

Usually that call comes too late. Passive candidates aren't in a hurry; that's what passive means. They answered because they were curious, or a bit restless where they are, or because your message happened to land on the right day. That mood doesn't last. If the recruiter calls four days later, the person on the other end has mostly moved on, and the recruiter is restarting a conversation that has already gone cold.

Where the bottleneck went

The pitch for AI sourcing rested on an assumption that sounded reasonable: the hard part of recruiting is finding the right people, and once they're found, recruiters can handle the rest.

That only works if recruiters have spare time to absorb whatever sourcing produces. They didn't have much before these tools arrived, and the tools multiplied the output.

Put rough numbers on it. Say a sourcing workflow used to turn up 30 to 50 qualified prospects a week for a role. A recruiter could follow up with that many alongside everything else. If the same workflow now surfaces 200 or 300, with messages already drafted and queued, the front of the funnel has been rebuilt while the part that holds conversations hasn't changed at all.

This is why response rate gets so much attention. It's the number you can see. But imagine you doubled it tomorrow. The recruiter still couldn't have 400 real conversations in a week. What limits you now is how many conversations your team can hold, and sourcing tools weren't built to help with that.

The best leads get the worst treatment

What makes this more than an efficiency problem is that the candidates stuck in that backlog are often the good ones. Sourcing found them because their background matched specific criteria. They didn't stumble onto your posting; you went looking for them. That's a different kind of lead from someone who applied to 40 jobs in a single evening.

Yet they get the worst experience in the funnel. An inbound applicant at least gets a confirmation email and eventually a call. A sourced candidate who wrote back to say they were interested often hears nothing for four days, then gets a follow-up that's obviously templated. When a recruiter finally calls on day six, the candidate is polite and vague, and the lead is gone.

So the finding is automated and the follow-up is still manual, limited by the hours recruiters have in a day. You've built a smarter funnel and pointed it at a team that already couldn't keep up with the leads it had. That gap is the difference between tools that assist and tools that act, which I wrote about in the post on agentic recruiting.

What would help is shrinking the time between "this person is interested" and "this person has had a real conversation about the job." That's the stretch of the process we built Asendia for. The limit there is a recruiter's day, which only holds so many calls, so interested replies wait behind everything else. Software that phones people doesn't have a queue like that. When a sourced candidate replies, Asendia can call them in the same session, evenings and weekends included, and hold the first screening conversation. The call is structured but adapts to what the candidate says. It checks them against the role's criteria, answers their questions about the job, and writes a detailed summary into the ATS for the recruiter.

The recruiter's day then looks different. Instead of chasing 150 leads that get a little colder every hour, they start from a ranked shortlist of people who have already talked about the role, with notes on fit, motivation and availability. For an agency running sourced outreach, every interested lead gets a first conversation within hours instead of days. I think this is part of why the recruiting agency model is splitting in two: the agencies that can absorb this volume are the ones pulling ahead.

AI sourcing did something impressive. It made finding the right people much easier and cheaper. It just didn't touch what comes after. If your sourcing numbers look great and your pipeline still feels thin, go look at the interested replies sitting in your outreach tool. The bottleneck didn't disappear when these tools arrived. It moved to the first conversation, where it's easy to miss.

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