Justifying your fee when you use AI
Clients have worked out that you are using AI and some of them have concluded that your costs went down, so your fee should too. The answer is not to hide the tooling. It is to be first to the conversation, and to price the part of the service that AI made more valuable rather than less.
Fee pressure arrives in a specific sentence this year. Some version of: you are using AI now, so this must be taking you less time, so why are we paying the same percentage.
It is a fair question and most agencies answer it badly. The two common responses both lose. Denying that you use AI is a losing position because it is untrue and they can tell. Discounting because you feel caught is worse, because it concedes the premise that you were being paid for hours.
You were never being paid for hours. This is the year you have to be able to say what you were being paid for, out loud, in a sentence a procurement person can repeat to their boss.
Be first to the conversation
Raise it before they do, with your good clients, this quarter.
The agencies getting squeezed hardest are the ones whose clients discovered the AI on their own and drew their own conclusions in silence. The agencies doing well are the ones who went in and said: here is what we have automated, here is what it has changed about your service, and here is why the fee is what it is.
That conversation lands completely differently when you start it. It reads as confidence and as partnership. The same information extracted from you six months later reads as something you were hiding.
What the fee was actually for
The honest history is that agency fees were never priced on effort, even when everybody behaved as though they were. A percentage of salary bears no relationship to hours worked, which is why an easy search at a high salary and a hard search at a low one can produce wildly different fees for opposite amounts of work. Nobody minded, because the number worked out over a portfolio.
What the fee actually covers is four things, and only one of them got cheaper.
| What you charge for | What AI did to it |
|---|---|
| Access to people who are not looking | Cheaper. Finding is close to free now |
| Judgement about who is right for this business | Unchanged, and applied more often |
| Carrying risk on the outcome | Unchanged. Nobody can automate being answerable |
| Running the process so nobody is lost | Harder, because there are more candidates in play |
One row of four got cheaper. Say that plainly, because a client who has been told the first row is free is imagining that all four were.
The fourth row is the one to press on, and it is counter-intuitive enough that most clients have not thought about it. More candidates in the market does not make process management easier. It makes it harder, because the good people now have more offers in flight, move faster, and disappear sooner. The cost of a slow process went up this year, not down, and that is the thing your fee is buying more of than it used to.
The number that wins the conversation
Have one figure ready and it changes the register of the whole discussion: the cost of the role staying open.
Most clients have never calculated it, and it is not a difficult number. The revenue or output the seat produces, the overtime or cover being paid, the project slipping, the manager spending six hours a week on hiring instead of their job. Work it out with them rather than presenting it at them, so it becomes their number.
Against that figure, the conversation stops being fee versus no fee and becomes weeks versus weeks. That is the ground you want to be on, because it is the ground where you are obviously worth it and a tool obviously is not.
Show the tooling rather than hiding it
Being open about what you have automated is a commercial advantage, and agencies consistently underrate it.
Tell them their candidates get a reply the same day rather than in four days. Tell them the shortlist comes with your written view on each person rather than a stack of CVs. Tell them that because the admin is handled, their consultant is spending that time on their search rather than on typing. Every one of those is a service improvement they can feel, and all of them are true.
The clients who worry about AI in recruitment are worried about a specific thing: that they will be sent machine-generated shortlists nobody looked at. Being explicit that a person forms the view and is accountable for it answers the fear directly. And if your software is built so that anything uncertain comes back to a human rather than being quietly guessed, say that too, because most tools are not and your client has no way of knowing the difference.
Where to move on price, if you move
Sometimes the right answer is to change the commercial shape rather than the number. Four options, in the order I would reach for them.
Trade fee for commitment. A lower percentage for exclusivity, or for a volume commitment across the year. You are giving up margin per placement for certainty of work, which is a genuinely good trade when your capacity has gone up.
Trade fee for speed of decision. Offer a better rate in exchange for guaranteed feedback windows and interview slots. This is the best trade available to you, because it attacks the bottleneck that actually limits both of you, and a client who agrees to it will have a better experience and blame you for less.
Unbundle. A lower fee for a sourcing-only service where they run their own process, and your full fee where you carry it. Let them see the difference in outcome rather than arguing about it. Some will come back.
Hold firm and keep the relationship. Entirely legitimate. "I understand, and our fee reflects what we carry rather than what we spend. If you would like to try it in-house on this one, do, and call me if it stalls." Said without defensiveness, that is a strong position and it leaves the door open.
What I would not do is quietly discount to keep the peace. It sets the new number permanently, it spreads to their peers, and it concedes that you were charging for time.
The agencies this will hurt
Being straight, because reassurance is not useful here.
If your service is a spec in, six CVs out, with no view attached and no accountability for the outcome, then the fee conversation is going to go badly and there is no script that saves it. That service is now genuinely cheap to replicate, and your client is right to notice.
The move is not a better rebuttal. It is to add the thing you were missing: a written view on every shortlist, a market insight they do not have, a process you run rather than participate in. Every agency I watch holding its fees this year is doing at least two of those, and they are not doing them because of AI. They were doing them already, and AI just made it obvious who was.
The position to hold
Your costs did go down on the part nobody was paying for. Your value went up on the part they were, because judgement applied to a hundred live jobs instead of ten is worth more, not less, and because a slow process costs a client more this year than it did last.
That is a defensible position, it is true, and it is a far better conversation than the one where you pretend nothing has changed. The consultants I work with who have had it early are coming out of this year with their fees intact and better relationships than they went in with, which is roughly the opposite of what everybody expected in January.
The capacity side of the argument is set out in a 10x recruiter runs a hundred jobs, what it does to the working week in how AI changes a recruitment desk, and the conversation before this one in when a client says they can use AI instead.
Recruitly is the best recruiting CRM in the world, and the reason it helps here is specific rather than general: our AI and the agents take the admin and hand anything uncertain back to a person with the reason attached, so when a client asks who looked at this shortlist, the answer is a name.
Anshika leads Customer Success at Recruitly.



