The Two Kinds of Recruiting Agency

What does a client actually pay a recruiting agency for? Some of it is judgment: knowing a market, knowing which candidates are good, knowing how to close them. But a lot of it, especially on high-volume work, is capacity. The client has more roles than its own team can handle, so it rents a team that can.

I think those two things are starting to be priced very differently, and that this will split agencies into two groups.

Look at it from the client's side first. Corporate HR teams are building their own AI tools. For the most part they aren't replacing their agencies, at least not yet, but they're quietly taking on more work before they need to call one. An in-house team that can see its own screening getting cheaper and its shortlists arriving faster starts asking harder questions about the capacity it pays for outside.

Imagine a CPO in a quarterly review asking why a retained agency charged $18,000 for a campaign that took three weeks and produced six qualified candidates. A few years ago that was just what hiring cost. Now the same person can watch their internal team's numbers improve month by month, while the agency's progress mostly arrives as "we're working on it." That is a harder comparison for the agency to win, and it doesn't take the agency doing a bad job. The best agencies are very good. But the gap between a strong human team and a similar team with AI doing part of the work has grown wider than a retainer can cover.

Plenty of agencies have adopted AI tools, but mostly in the obvious places: writing job ads, helping with outreach, scoring resumes. Those help a little. They don't change how much work the team can take on, because none of them touch the step that limits it.

The step that limits it is the first conversation. Say a recruiter can do about ten good screening calls a day, with nothing else competing for their time. Over a month that's around two hundred people. A single high-volume campaign for a retail or logistics client can bring in a few hundred applications in its first three days, before the recruiter has looked at any of them. So the agencies that sped up their job ads and outreach are a bit faster at the top of the funnel, and still stuck at exactly the point where applicants pile up. That backlog is usually where client dissatisfaction comes from.

A smaller group of agencies has started handling first contact differently. They no longer assume a human has to make the first screening call. AI handles the volume there, and recruiters spend their time where judgment matters. What changes is that the process keeps running when the recruiters don't. An application that comes in late on a Friday can be screened before Monday morning, and a client who used to wait until Wednesday for a first shortlist can have one on Monday afternoon. Every applicant gets a real conversation within hours instead of days. Differences like that come up when clients decide whether to renew.

The ten-calls-a-day ceiling is the thing we built Asendia to lift. A recruiter can only hold so many first conversations before the day runs out, but software can talk to every applicant in a campaign, so nobody sits in a queue waiting for a free slot on someone's calendar. When a person applies, it calls them that evening or that weekend and runs a structured qualification screen against the criteria set for that role. The follow-up questions depend on what the candidate says, so you learn how they think and talk rather than how well they can polish a written answer. What goes back into the ATS is a qualification summary, key quotes and a ranked shortlist, so on Monday the recruiter starts with a few dozen people who have already been talked to instead of hundreds of unread applications.

The interesting consequence is economic. If the AI absorbs volume spikes, the agency no longer has to hire for its busiest weeks. The team stays the same size and its output can go up or down with demand. That looks more like the margins of a software company than a services business, which is an odd thing for an agency to have and probably a lasting advantage for the ones that get it. We've written more about why AI that actually runs steps of the pipeline beats AI that only assists, in our post on agentic recruiting.

None of this means agencies are going away. Relationships, market knowledge and specialist placement skill still matter a great deal, and AI doesn't supply them. What's changing is the cost of the capacity that surrounds them. An agency that can take in five hundred applications over a weekend and one that can't are going to look more and more different to clients, and I don't expect that gap to close by itself.

If you run an agency, one number will tell you which group you're in. Take your last high-volume campaign and count how many applicants had a real conversation with someone within a day of applying.

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