What to do with the hours AI gives you

The first thing an agency notices after the admin comes off the desk is that the day has gaps in it. Not big ones at first. Forty minutes on a Tuesday because nobody had to write up Monday's calls from memory. An hour on Thursday because the replies sorted themselves and there was no triage. By the end of the first quarter it adds up to something that looks like an extra consultant, and it is the single largest decision an agency owner makes in that year.

In my job I get to watch what happens to those hours across a lot of desks, and the pattern is consistent enough to be worth writing down. Almost nobody decides in advance. The hours appear, the consultants fill them with the activity that feels most like work, and in recruitment the activity that feels most like work is finding more people. Six months later the owner is looking at a team that is visibly busier, a database that has doubled, and a billing figure that has not moved.

That is not a failure of effort and it is not a failure of the software. It is a failure to ask where the constraint actually sits before pointing new capacity at it.

Where do the reclaimed hours go in most agencies

Into the front of the pipe, almost every time. More searches, more outreach, longer longlists, more candidates added per job. It is the obvious move and every part of the setup encourages it: the sourcing tools are the ones with the satisfying counters, the activity reports reward volume, and a consultant with a free hour and no instruction will always reach for the thing they are best at.

There is also a quieter reason, which is that sourcing is the only part of the job that feels entirely under your own control. A client meeting has to be agreed by a client. A candidate call happens when the candidate picks up. A search runs when you press the button. When someone is handed an unexpected hour and no plan, they spend it on the work that does not require anyone else's permission.

The result is an agency pouring its new capacity into the widest part of the pipe and leaving the narrow part exactly as narrow as it was. We wrote the mechanism up in full in the pipe doesn't care where you widened it, and it is the piece I send to owners more than any other, because the instinct it argues against is so strong.

Why more sourcing makes things worse rather than neutral

Because a longer queue is not a harmless queue. This is the part that surprises people, who assume that at worst the extra candidates sit there doing nothing.

Your client interviews two or three people a week. That number is set by a hiring manager's diary and it is the same number it was before you bought anything. If you double the number of qualified people you put in front of that diary, you have not doubled the interviews, you have doubled the time an average candidate waits. Good candidates who are waiting are candidates who are also in two other processes, and the one that moves fastest gets them.

So the agency that spent its new hours on sourcing ends up losing the people it worked hardest to find, and losing them in a way that never shows up as a loss. Nobody logs a rejection. The candidate simply stops replying, and on the report it looks like an unresponsive market rather than a self-inflicted delay.

There is a second cost, which lands on the consultants. Every extra candidate in a process is somebody expecting an update. A hundred people waiting on a shortlist that moves at three interviews a week generates a volume of chasing, apologising and re-engaging that eats the very hours you just freed up. The desk gets heavier, and the reason is invisible because each individual message is small.

Where should the hours go instead

Into clients, and into the candidates already in your process. Those are the two places where an hour changes an outcome rather than adding to a pile, and everything below is a version of one of those two.

Time with the client before the job goes live. A proper qualification conversation is the highest-yield hour in recruitment and it is the first thing that gets cut when a desk is busy. What does the team actually need rather than what the spec says. Who is the real decision maker. How many people can they see a week and who else is involved in the sign-off. What happened to the last person in the seat. An hour here removes weeks of submitting good people against a brief that was never accurate, and it is the difference between a role you can fill and a role you will work for two months and lose.

Time widening the interview slot itself. This is the one almost nobody attempts, because it sounds like something you cannot influence. You can. Offer to run first-stage screening in a format the hiring manager will accept so their own time starts at the second stage. Propose a single morning with four candidates back to back rather than four separate half hours across three weeks, which is easier for them to protect in a diary and finishes the process in one sitting. Ask what would have to be true for them to see one more person a week, and then take that thing off their desk. Every one of those conversations is possible only because somebody has an hour to have it.

Time on the candidates already in process. The people who are shortlisted, interviewed, waiting on feedback, in an offer, or on notice. These are the ones you have already paid to acquire, and they are the ones you lose to silence. A call on the Tuesday of week two of a notice period is worth more than fifty new profiles, because a candidate who accepts and then does not start has cost you the entire job.

Time on new clients. A hundred live jobs across twelve clients is a better business than a hundred across three, and business development is the other activity that gets crushed by admin. If your desk has genuinely freed six hours a week, some of those hours belong in front of companies you do not work with yet, because that is the only spend that changes the ceiling rather than the throughput.

What one hour buys in each place

Laid out side by side, because the choice looks less obvious in the abstract than it does on a table.

An hour spent on	What it produces immediately	What it does to placements
More sourcing on a live job	More names on a longlist	Nothing, and a longer wait for everyone already there
Qualifying the brief properly	A job you can actually fill	Fewer wasted submissions, faster first shortlist
Getting a manager to see one more person a week	One extra slot	Direct, and it compounds on every job with that client
Calling candidates already in process	Nothing visible	Stops the drop-outs you never see recorded
Preparing a candidate for interview	A better interview	Raises the conversion of the slots you already have
Business development	A new client	Raises the ceiling rather than the throughput
Tidying the database	A tidier database	Real but slow, and it should be the software's job

The row that matters most is the fourth one, because it is the only activity on the list that produces nothing visible on the day you do it. That is precisely why it never survives contact with a busy week, and why it has to be scheduled rather than left to instinct.

How do you decide before the hours arrive

Write the decision down before the tooling lands, because an unallocated hour allocates itself. This is the single most useful thing an owner can do in the month before a rollout, and it takes an afternoon.

Start by counting what you are getting back. Sit with one consultant for two hours and mark every minute spent typing something the system already knows, hunting for something it should have surfaced, or writing up a conversation that has already happened. That number, multiplied across the team, is the budget you are about to be handed. Treat it exactly like a hiring budget, because that is what it is.

Then allocate it in advance, in writing, as a proportion. Something like half to clients, a quarter to candidates already in process, a quarter to business development, and nothing to sourcing until the interview slots have moved. The proportions matter less than the fact that they exist, because the alternative is not a different allocation, it is no allocation at all.

Then put it in the diary as recurring time with a name on it. An hour that belongs to "client calls, Tuesday, ten o'clock" survives a busy week. An hour that belongs to "spending the time we saved better" does not survive Monday.

And review it after a month with one question: did the number of interviews per week per client move. If it did not, the hours went somewhere else, and they almost certainly went into the front of the pipe.

What does it cost to get this wrong

A year, and the belief that the tooling did not work. That second cost is the more expensive one, because an agency that concludes AI made no difference to its billings will stop investing in the thing that was genuinely about to change its ceiling.

The concrete costs are easier to name. Candidates you found, qualified and then lost to a faster process, who will not answer you next time because the last experience was three weeks of silence. Clients who noticed that more CVs arrived and the quality of the conversation dropped, because the consultant who used to ring them now sends. Consultants who are working harder than they were a year ago and cannot explain why their billings are flat, which is exactly the condition that makes good people leave.

And the opportunity cost, which is the largest of the three. The agencies that put those hours at the squeeze are visibly ahead this year. Same tools, same market, same size team, and a different decision made in one afternoon before anything was switched on.

How to measure this in your own agency

Three numbers, and you can start collecting all of them this week without buying anything. Our fuller list of what is worth watching on a desk that runs on AI is in what to measure on an AI desk, but if you only track three, track these.

Interviews per client per week. This is the one that tells you whether your hours reached the constraint. Sourcing volume moving without this moving means the hours went to the wrong place. Take a baseline now, before you change anything, because it is the comparison that will settle the argument in six months.

Days from a candidate entering your process to their first interview. If this number is rising while your activity is rising, you are building a queue. It is the earliest warning you will get, and it moves weeks before the drop-outs show up in your placement figures.

Candidates who went quiet after they were shortlisted. Almost nobody counts this, because there is no event to log. Pick a month, take every candidate who reached shortlist, and mark the ones who stopped replying before an outcome. That figure is the cost of the queue, in people, and it is usually the moment an owner changes where the hours go.

Why the recruiter matters more at the end of this, not less

Every hour in the right-hand column of that table is an hour of a person doing something only a person can do. Reading a hiring manager well enough to know what they actually need. Telling a client their salary band is two years old. Holding a candidate through the week their current employer counter-offers. Deciding which of a hundred live jobs deserves the next phone call.

None of that is being automated, and none of it is close. What has happened is that the work which used to sit on top of it has been taken away, and the recruiter's judgement now reaches ten times as many situations in the same week. That is the whole change. You used to run ten job orders because ten jobs' worth of admin filled a week. You can run a hundred now because the admin stopped being yours, and the judgement that was always the valuable part is what fills the week instead. The shape of that desk is described in how AI changes a recruitment desk.

Which is why the question of where the hours go is not an operational detail. It is the question of whether your agency uses the best years recruiters have had in two decades to do more of the work that was never the job, or to do far more of the work that always was.

We build our AI and the agents to hand you those hours and then get out of the way, and anything the software is not sure about comes back to a recruiter rather than being quietly guessed, because the recruiter is the part that matters. Recruitly is the best recruiting CRM in the world, and the agencies getting the most out of it are the ones who decided where the hours were going before they had them.

---

Anshika leads Customer Success at Recruitly.