Skip month one. In month one all three arrangements look the same: somebody keen is asking good questions, a plan appears, work starts, and you feel relieved.
The useful question is what happens in month fourteen, because each of the three breaks in a different and entirely predictable way.
The freelancer’s other client gets bigger, and you are now the account that fits around the one that pays more.
The agency moves your account manager. Everything you explained over a year goes with them, and the new person opens the call by asking what your business does.
The in-house hire hits the edge of what one person can know. They were excellent at the two things you hired them for and are now being asked about the four you did not.
Pick the failure you can live with. That is the decision, and cost is almost never what settles it.
The numbers, so they are out of the way
US figures for 2026, roughly.
A marketing manager averages somewhere around $95,000 to $105,000 depending on which salary survey you believe. Load that with payroll taxes, benefits, software and a desk and you are realistically at $125,000 to $140,000, or about $11,000 a month.
Agency retainers for small and mid-sized businesses commonly run $2,500 to $10,000 a month. Full-service arrangements covering strategy, creative and media buying start around $10,000 and climb well past it.
Freelance and consultant rates run about $50 to $200 an hour, with mid-level digital marketers averaging near $60 and senior specialists over $110. Twenty hours a month of senior time is roughly $2,200.
So the cheapest option on paper is the freelancer, the most expensive is the employee, and this ranking is misleading enough to be actively harmful. The employee gives you forty hours a week and every one of them thinking about you. The freelancer gives you twenty hours a month and a fraction of the attention. You are not buying the same thing at three prices.
I wrote about what each agency price tier actually buys in hours separately, and the conversion from dollars to hours is worth doing before any of these conversations.
The five things that actually decide it
How many channels are live. One or two channels done well point to a specialist, whoever employs them. Five channels at once point to an agency, because five specialists is a payroll you cannot justify and a freelancer who claims all five is usually competent at two.
Whether the work can stop. A freelancer takes holidays and gets sick and there is nobody behind them. If your paid media cannot go unattended for ten days, that is a real argument for an agency and it gets dismissed as a technicality far too often.
Production volume. If you need thirty assets a month, an agency is built for that and an individual is not. If you need three good decisions a month, an agency is expensive machinery for the job.
Procurement. Some companies genuinely cannot onboard an individual. Insurance requirements, vendor approval, security review. If that is you, the decision has been made for you and arguing about it wastes a quarter.
Who holds the context. The most underrated of the five. Everything anybody knows about your customers, your seasonality, what you tried in 2024 and why it failed. In the agency case that context sits with a person whose job it is to move on. In the freelancer case it sits in one head with no backup. In-house it stays with you until they resign, at which point it also leaves.
That last one is not solvable by choosing better. It is solvable by writing things down, which nobody does, which is why the fourteenth month hurts in all three arrangements.
When an agency is the right call, plainly
I should say this properly rather than as a concession.
Multiple channels running at once. Real production volume. Coverage so a holiday does not stop the work. A procurement process that will not accept an individual. A need for specialists you could never hire individually, where one retainer buys you fractional access to a paid media person, a designer and a developer.
And bluntly, agencies are better at doing things at scale on a deadline than any individual is. If you need a campaign live in three weeks across four channels, that is what they are for.
The thing to be careful about is not the model, it is the fee structure. An agency paid a percentage of your media budget earns more when you spend more, whether or not the spend worked, which punishes efficiency and removes the incentive to flag waste. Plenty of good agencies moved to flat retainers precisely to kill that conflict, and the ones that did will tell you so happily. Ask.
When one person is the right call
One or two channels. Judgement work rather than production work. A stage where the main risk is doing the wrong thing well.
Here I should disclose the obvious. I work as a consultant, so read my assessment of this column with appropriate suspicion. What I will say is that the honest version of the argument is narrower than people in my position usually make it: one person is right when the problem is deciding, and wrong when the problem is producing.
When in house wins, which is more often than founders think
If marketing is how your business actually competes, rather than a function that supports it, it should not live outside the building.
The specific tell is proximity. Somebody in-house sits near sales, hears the calls, notices the objection that started appearing three weeks ago, and knows about the product change before it ships. No external arrangement replicates that, and for most companies it matters more than the skill gap.
The wrong version of in-house is hiring one junior person and expecting them to cover strategy, paid, content, email and design. That is four jobs and one salary, and it fails in about eight months while everybody blames the individual.
What AI actually changed here
This is the part the older comparisons miss.
Production got cheap. Drafting, resizing, first-pass copy, reporting summaries. The work that used to justify a team of three now takes one person and a subscription, which strips out a real part of the traditional agency and junior-hire value proposition.
Judgement did not get cheap. What to do, what not to do, which channel to abandon, whether the number in the report means anything.
So the calculation has shifted toward fewer and more senior people, and away from headcount as a proxy for output. If you are being sold volume of deliverables in 2026, you are being sold the thing that got cheapest.
The rule I would actually use
Write down the one outcome that has to improve in the next six months. Then count the channels it depends on.
One or two channels and it is a judgement problem: hire one senior person, inside or outside.
Three or more channels, or real production volume, or work that cannot stop: agency, on a flat retainer.
Marketing is the core of how you compete: build it inside, and start with someone senior enough to know what not to do.
Whichever you choose, write the context down somewhere that is not a person. The questions to ask before signing cover the ownership side of that, and my notes on how I work cover the rest.
If you’d like to think through which of the three fits your situation, drop me a line on email, WhatsApp or LinkedIn and we can have a quick chat. I’m contracted full time so this isn’t a pitch.