If you have opened a job description recently and felt a quiet unease about whether you are even hiring for the right thing anymore, you are not alone. Figuring out how to hire marketing staff in an AI world is one of the more disorienting problems facing agency owners right now, because the tools have moved faster than the org chart. The skills that justified a salary twelve months ago may now sit inside a platform. The skills that actually matter are harder to name, harder to test for, and almost never listed on a CV.
This is not a piece about replacing people. It is about making sure that when you do hire, you are not paying a premium for capabilities the agent already covers, and that you are genuinely rewarding the judgment that no tool can replicate.
The job description you are still using was written for a different agency
Most agency job specs for content and marketing roles were written, or at least shaped, during a period when output was the constraint. You needed people who could produce: write briefs, draft copy, build decks, populate calendars. Volume was the problem, and headcount was the solution.
That is no longer the shape of the problem for agencies running a proper marketing platform. Output has become largely a solved problem. A well-configured agent, trained on your clients’ brand memory and past deliverables, can draft, adapt and schedule across channels at a pace no team of writers could match. The constraint has shifted from production to judgment, and most job descriptions have not caught up with that shift.
The result is that agencies are interviewing for the wrong signals. Word count per hour. Familiarity with content management systems. A portfolio of articles. These things are not worthless, but they are no longer the load-bearing skills they were. Hiring on those criteria now is a little like hiring a fast typist because you need a strategist.
What you are probably overpaying for
Raw content production speed
A candidate who leads with output volume as their value proposition is telling you something useful: they have optimised for the old constraint. That is not a character flaw, but it is a mismatch. If your agency is running campaigns through a platform that handles first drafts, channel adaptation and scheduling, then paying a senior salary for someone whose primary skill is writing fast means you are funding duplication. The platform already does that work. What you need is someone who can review, direct and improve what it produces.
Broad channel familiarity
Knowing how Instagram captions differ from LinkedIn posts, or how email subject lines behave, used to be a genuine differentiator. It is still useful context, but it is no longer rare. Any platform worth using has that knowledge baked in, and any reasonably curious person can acquire it quickly. Hiring as though multi-channel fluency is a scarce skill means you are competing for something that has commoditised.
Tool proficiency as a headline skill
Listing ten software platforms in a CV is not the same as demonstrating judgment. Tool familiarity fades fast; the tools change every few months. Someone who has used four different project management systems probably adapts well, but that adaptability is the skill worth paying for, not the familiarity itself.
What actually matters now, and how to test for it
Strategic literacy
Can this person read a client’s commercial situation and form a view about what the marketing should be doing? Not just what content to produce, but what problem the content is solving, for whom, and why it should work. This is the skill that sits upstream of everything the agent produces. If no one in the process can make that judgment, the output will be competent but directionless.
You can test for this in an interview without a formal case study. Describe a real client situation in general terms and ask the candidate what questions they would want answered before recommending a channel strategy. A strong candidate narrows down to the commercially significant questions quickly. A weaker one reaches for tactics before they have understood the problem.
Editorial judgment, not copywriting craft
The distinction here matters. Copywriting craft is about producing polished sentences from scratch. Editorial judgment is about reading output and knowing whether it serves the brief, whether the tone is right, whether the argument holds, and what needs to change. One of those skills is more in demand than it has ever been. The other is increasingly assisted.
Ask candidates to review a piece of draft content rather than write one from scratch. Give them something that is technically competent but subtly wrong: the wrong register for the audience, or a structure that buries the point. The people you want to hire will tell you specifically what is wrong and why. The people who are less useful will tell you it is good with a few minor tweaks.
Client-side communication and the ability to manage approvals
Getting campaigns approved, keeping clients confident and translating feedback into something actionable is genuinely difficult. It requires reading people, managing expectations and holding a position when the client’s instinct conflicts with the strategy. No platform automates this. The approval workflow might sit inside your tools, but the relationship sits with your team.
This is often underweighted in hiring because it is hard to assess from a portfolio. Look for candidates who can describe a moment when a client pushed back on something they believed in and explain what they did. You are not looking for stubbornness or compliance; you are looking for someone who knows when to hold and when to adapt, and who can articulate the difference.
Curiosity about how the tools work, without being intimidated by them
You do not need every hire to be technically deep. But you do need people who are willing to understand what the agent is doing, why it produced a particular output, and how to improve the instructions that shape it. Someone who treats the platform as a black box and accepts whatever it produces uncritically is not adding the layer of judgment you are paying for. Someone who interrogates the output and gets better results from the same tools over time is compounding in value.
How to hire marketing staff in an AI world when the role itself is uncertain
Sometimes the honest answer is that you do not yet know what the role should look like, because you have not fully defined how the platform fits into your delivery process. Hiring into that ambiguity tends to go badly: the new person fills the vacuum with what they already know how to do, which may not be what you need.
Before you write the job spec, it is worth mapping what the agent currently handles in your workflow, what still requires human judgment, and where the gaps are creating quality or capacity problems. That exercise usually clarifies whether you need a strategist, an account lead, a production manager, or a hybrid that does not have a clean job title yet. The answer shapes everything: the salary, the seniority, the interview questions, the onboarding.
It also helps to be honest in the job description itself. Candidates who are thoughtful about AI tend to respect transparency about how you are using it. Describing the role accurately, including that the agent handles first drafts and the human handles direction, approval and client communication, will attract people who are comfortable with that dynamic. It will put off people who want to spend their days writing, which is probably the right filter.
A note on junior roles specifically
The question of whether to hire junior writers at all is one that deserves its own honest conversation. If the platform is producing drafts that your senior team reviews and approves, then a junior writer whose main contribution is producing more drafts may not have a clear place in the workflow. The more useful junior role is probably one oriented around coordination, quality review, client-facing support and learning the strategy side of the work. That is a different profile, and it is worth being deliberate about it rather than hiring the familiar shape and hoping it fits.
There is a longer version of this conversation in the piece on whether agencies should replace junior writers with AI, which covers the human side of that decision as well as the operational one.
Retaining the people you already have
Hiring decisions do not happen in isolation. Your existing team is watching how you are handling AI, and the way you talk about new hires signals something about how you value the roles already in the building. If the people you bring in are clearly positioned as AI operators and editors rather than creators, your existing team will draw their own conclusions about where their career is heading.
The agencies that handle this well are usually the ones that involve their current team in defining how the platform fits into the workflow, rather than presenting it as a fait accompli. The skills your current people have built around strategy, client relationships and editorial judgment are exactly the skills that matter most now. Framing that honestly, and investing in developing those skills rather than replacing the people who have them, tends to produce both better retention and better work.
If you are thinking through how to have that conversation with your team, the piece on what your best people are thinking about AI covers the dynamics that agency owners often underestimate.
One hiring principle worth keeping
Hire for the judgment the platform cannot replicate, not for the output it already produces. Every time you write a job spec, ask yourself what this person will do in the moments when the agent produces something that is technically correct but strategically wrong. If the answer is not clear, the role is not yet defined clearly enough to hire for.
Verbolica is built around the idea that the agent handles the operational weight of running campaigns, so your team can focus on the thinking that shapes them. If you want to see how that changes the structure of a delivery team in practice, the features overview shows what the platform takes on and where the human role sits in the workflow.
