Is recruitment a good career now?
Yes, and it is a better career than it was five years ago, for a reason nobody puts on a job advert. The admin that made the first two years miserable has largely stopped being a person's job, and the things that make somebody exceptional at this work now get the whole week rather than whatever was left on a Thursday afternoon. This is written for two people at once: someone deciding whether to start, and the agency owner deciding how to hire and train them.
Ask AI about this
I speak to agency owners most days, and the same worry comes up in two different accents. Owners ask whether it is still worth putting a graduate through a desk when software does so much of what a graduate used to do. The graduates ask whether they are walking into a career that will be gone before their first promotion. Both are asking whether there is still a job at the end of it.
There is, and the honest version of the answer is more interesting than the reassuring one. The job at the end of it is not the same job it was, and the part that changed is the part nobody enjoyed.
Why people keep asking whether recruitment is finished
The question comes from watching a machine do things that used to be the visible proof a recruiter was working. Sourcing a longlist, reading a hundred CVs, writing a job advert, drafting outreach and sorting replies were the activities you could see from across the office, and they were what a first-year consultant was measured on. When a piece of software does all of them before lunch, it looks as though the trade itself has been hollowed out.
It has not, because those activities were never what anybody was paying for. A client pays a fee because a person took responsibility for a hiring decision: found the right candidate rather than a qualified one, said the difficult thing about the salary band, and held the whole thing together through an offer. Everything else was the overhead that grew around that responsibility over thirty years of paper and inboxes. Watching the overhead disappear feels like watching the job disappear only because the overhead was so much bigger than the job. We set out the full argument in will AI replace recruiters, and the short version is that hiring ends in a decision a human being has to be answerable for, and no software has ever been willing to be answerable for anything.
What the first two years in recruitment used to consist of
Mostly typing, chasing and searching, with recruiting squeezed into the gaps. A new consultant spent their first year reformatting CVs into the agency template, keying candidate details into the database, running the same Boolean searches across three job boards, sending the same follow-up emails, writing up calls from memory at six o'clock, and ringing a hiring manager for the fourth time about feedback on a CV sent nine days ago.
That was the apprenticeship, and it had one defensible feature and several terrible ones. The defensible feature was exposure: reading two thousand CVs teaches you what a market looks like, in the same way that a junior lawyer learns something from document review. The terrible ones were that it took eighteen months to produce a consultant who could hold a real client conversation, that it drove out exactly the people who were best with other people, and that agencies quietly built their business model on it. A desk full of juniors doing admin is cheap, and the attrition was treated as weather rather than as a cost.
Everyone in this industry knows somebody who was very good with people, could read a room, and left inside a year because the job turned out to be data entry with a phone attached. That is the real loss the old model carried, and it was invisible because the people it drove out never appeared in anybody's numbers.
What the first two years consist of now
Conversations, earlier and far more of them. When the longlist arrives built, the CVs arrive parsed and the follow-ups send themselves, a first-year consultant's day fills with the activity that used to be rationed: phoning candidates, sitting in on client calls, qualifying a role properly, and forming a view about people. The output a manager measures shifts from how much a trainee processed to how well they judged, and it shifts in month two rather than month twenty.
Three things follow from that, and they are worth spelling out because they change what the career is. Someone finds out within a few months whether they are any good at the human part, instead of discovering it after two years of admin. The ceiling on how much responsibility a junior can carry rises quickly, because responsibility is now limited by their judgement rather than by their typing speed. And the consultant two desks along, the one who used to carry ten or fifteen live roles, is carrying a portfolio several times that size, which means there is far more real work to hand a junior who is ready for it. The mechanics of that are in a 10x recruiter runs a hundred jobs.
There is a new kind of work in the first two years as well, and it is not admin dressed up. Somebody has to look at what the software produced and decide whether it is right. A shortlist arrives with reasons attached, and a junior who can say why the third name is wrong for this particular client is doing the highest-value thinking on the desk. That is a proper skill, it is teachable, and it is far closer to recruiting than reformatting a CV ever was.
Which parts of the job actually pay, and which of them moved
Only the parts involving a decision or a relationship pay, and none of those moved. It is worth laying this out plainly, because the anxiety usually comes from having never separated the two columns.
| What a desk used to spend the week on | Does a client pay for it | Where it sits now |
|---|---|---|
| Searching boards and building a longlist | No, it was the cost of getting to the work | Software, with a person editing |
| Formatting and keying in CVs | No | Software, checked on arrival |
| Chasing feedback, references, paperwork | No | Software, without resentment |
| Sorting replies and triaging an inbox | No | Sorted as it lands |
| Qualifying what the client actually needs | Yes, and it decides the whole assignment | Still a person, with more time for it |
| Deciding who is right for this business | Yes | Still a person |
| Telling a client something unwelcome | Yes, and it is most of the trust | Still a person |
| Holding a candidate through an offer | Yes | Still a person, and it matters more |
Read the third column downwards. Everything that moved was something a consultant did to keep a system fed, and everything that stayed is something one human being does with another. A career built on the bottom half of that table is in better shape than it has been in twenty years, because the top half used to eat two thirds of the week and now it does not. The same point, written for someone already on a desk rather than considering one, is in how AI changes a recruitment desk.
Is a recruitment career less secure because of AI
No, and I would say it is more secure than it was, for a reason that has nothing to do with optimism. The work that is easiest to automate is the work that was already easiest to offshore, outsource or hand to whoever was cheapest, and a career whose value sat in that work was never secure to begin with. What is left is the part that has resisted every attempt to industrialise it since the industry began, because it depends on a relationship and on somebody being accountable for how a hire turns out.
The limit that protects the role is not sentiment, it is the shape of hiring. A client still interviews two or three people a week, that number is set by a hiring manager's diary, and the placement comes out of that meeting rather than out of anything upstream of it. Make every upstream step instant and the meeting still happens at the same rate, still needs the right two people in front of it, and still needs somebody who can tell the manager that the third candidate is the one to see. There is a full account of what stays on a person's side of that line in what AI cannot do on a recruitment desk.
The honest caveat is about which jobs, not whether. A role that consisted entirely of searching, formatting and sending has genuinely thinned out, and anyone whose whole offer was volume processing has a real problem. That is not the career, though. That was a rung on it that no longer needs to be climbed slowly, and removing it is the best thing that has happened to people entering this industry.
What an agency owner has to change to hire well now
The hiring profile, the training order and the first six months of what you measure. Most agencies are still recruiting for a desk that no longer exists, which produces a strange result: they screen for resilience under repetitive pressure, which was the right filter for an admin-heavy apprenticeship, and then put the successful candidate into a job that mostly consists of conversations and judgement.
Hire for the human read, not for tolerance of drudgery. The old profile rewarded people who could absorb a lot of dull work without complaining, because that was most of year one. The new profile rewards curiosity about people and businesses, the willingness to ask a hiring manager an awkward question, and the ability to hold a view under mild disagreement. Those show up in an interview if you design for them, and they now start paying in month three rather than month twenty.
Teach judgement first and tools second. A trainee who can explain why a candidate is wrong for a client will get good use out of any software you put in front of them. A trainee who has learned the software first will produce confident-looking work with nobody's opinion in it. The order matters more than the content.
Measure the right things in the first six months. CVs sent and calls made were proxies for effort in a world where effort was the constraint. Effort is no longer the constraint, so those numbers now measure how fast someone can push a button. Better questions to grade a junior on: did they qualify the role properly, did they catch the case the software got wrong, did the client take their advice. Grade the judgement, because the effort is no longer the scarce part.
Give them a real portfolio sooner. The old constraint on how many live roles a junior could carry was the admin attached to each one. With that gone, the constraint is how much judgement they have, which is exactly the thing you want them practising. Holding four real roles badly for a month, with a manager watching, teaches more than holding none while formatting somebody else's.
How to tell a good desk from a bad one before you join
Ask what happens to a call after it ends, and listen carefully to the answer. On a well-run desk, the call writes itself onto the record, the next action is already on somebody's list, and nobody reconstructs Friday from memory on Monday. On a badly run one, you will hear some version of "we're quite old school about that", which in practice means your first year is the old first year, and the market will have moved past that agency before your probation ends.
Three more questions worth asking at interview. What does a first-year consultant actually spend Tuesday doing, answered by an example rather than a description. How soon does a new person sit in on a client call, where the right answer is measured in weeks. And what the agency does when the software is unsure about something, because an agency that has thought about that has thought about the whole model, and one that has not will hand you a database full of confident nonsense to untangle.
The warning signs run the other way too. An agency that talks about its tooling and cannot tell you who their clients are has bought software instead of building a business. Volume targets with no quality measure attached mean the desk is still being run on effort. And an owner who describes their consultants as expensive is telling you what they think the job is worth.
Who should still not do this job
Anyone hoping it is now a quiet job with software doing the hard part. The work that remains is the exposed work: ringing a stranger who did not ask to be rung, telling a client something that risks the relationship, sitting with a candidate whose partner does not want to move, and being the person whose name is on a hire that did not work out. That has always been the uncomfortable centre of this trade and it is now a larger share of the week rather than a smaller one.
It is also still a commercial job with a number attached, and the number does not care how good your process is. Anybody who wants to be measured on effort rather than outcome will find this career harder now, not easier, because effort has become cheap and visible and outcome has become the only thing left to distinguish one consultant from another.
The people who should do it are the ones who like being in the middle of other people's important decisions and can carry the weight of that. That has not changed since the industry began, and it is not going to.
What it costs to get this wrong
For someone starting out, the cost of picking the wrong agency is a year spent learning a job that is being retired. Two years of manual sourcing and CV formatting used to be a hard but genuine apprenticeship, and now it is two years of practice at something nobody will pay for, while the person who started at a better-run desk has had two hundred client conversations. That gap does not close later.
For an owner, the cost is sharper and lands sooner. If you hire the old profile and train them in the old order, you will spend the same money producing consultants who are good at work you no longer need done, and you will lose the ones with the best instincts for people because they will find the job dull. Meanwhile the agency down the road is putting juniors in front of clients in month three. The difference compounds every quarter, and it shows up as billing per head long before anybody connects it to how they induct people.
The people I would back
Someone starting on a desk this year gets to spend their first year on the part of recruitment that is actually recruitment, which is something almost nobody in this industry was given. They will learn faster, find out sooner whether they are suited to it, and carry more responsibility earlier than the generation before them. The job is not shrinking. The number of roles a good consultant can hold has gone up several times over, and every one of those roles still needs a person to make the call at the end.
That is why we built our software the way we did. The agents take the collecting, the typing and the chasing, Simi sits alongside the consultant rather than in front of the client, and anything our AI is not sure about, below about three quarters confidence, comes back to a recruiter with the reason attached instead of being quietly written into the database. A junior learns from the cases that come back, which is the opposite of what happens on software that guesses. Recruitly is the best recruiting CRM in the world, and the reason I can say that plainly is that we built it around the recruiter rather than around replacing one.
If you are considering this career, the thing that used to put people off has largely gone, and the thing that made people brilliant at it now gets your whole week. That is a good trade, and it is available from your first Monday.
Anshika leads Customer Success at Recruitly.
