There are four ways agencies charge for the same work: a percentage of your ad spend, a flat monthly retainer, a fee per project, or a commission on the growth. What you end up paying depends on which one you sign far more than on the rate inside it, because each model pays the agency to do something different, and that does not change after the contract is signed. Ask which model first. The price is a negotiation. The incentive is structural.
How much does it cost to hire a marketing agency?
Less than you think for advice and more than you think for delivery, and any number quoted outside those two words is guesswork. Scope is the variable that moves the total, not the agency's hourly worth: how many channels are actually run rather than watched, whether anybody builds the pages and the flows or only recommends them, how often somebody looks at the account, and whether the reporting is a dashboard you were given a login to or a person writing down what they decided and why. Two agencies quoting the same figure can be selling work that differs by an order of magnitude, which is why the figure on its own tells you nothing.
Why does no agency publish a price?
Partly because scope really does vary that much, and partly because a published number is a position you have to defend before you have heard anything about the account. Both reasons are real. The useful response is not to hunt for the number anyway, it is to ask for the model and the scope in writing, then compare those. A written scope makes two quotes comparable. Two numbers without scopes are not comparable at all, however precise they look.
What each pricing model pays for
| Model | The agency earns more when | So it optimises for |
|---|---|---|
| Percentage of ad spend | You spend more | Budget growth |
| Flat monthly retainer | Nothing changes | Keeping the account |
| Fee per project | You buy another project | Finishing and reselling |
| Commission on growth | Your revenue grows | Revenue, ads or not |
What is wrong with a percentage of ad spend?
It pays the agency more when it spends more of your money, and that conflict shows up at the exact moment it matters most. Every account eventually reaches a point where the honest decision is to spend less: a campaign that has stopped converting, a season that is over, a product that should be paused. Under a percentage of spend, the honest decision is the one that costs the agency income. Most people in the industry are decent and will make the right call anyway, which is precisely why the model survives. It should not need decency to work.
The second problem is quieter. A percentage of spend prices your account by budget rather than by difficulty, so a large simple account subsidises a small complicated one, and the agency is paid least where the work is hardest. That is backwards, and it is usually the reason a smaller brand feels ignored.
What is wrong with a flat retainer?
It is the honest default and it still has a flaw worth naming: the fee is the same whether the month was transformative or quiet. There is nothing in the structure that pays for the extra push, and nothing that punishes a slow month either. What holds a retainer honest is not the model, it is whether the agency writes down what it did and what it decided, every week, in a form you can read without a login. A retainer with real reporting is fine. A retainer with a dashboard and a quarterly call is a subscription.
When is a fee per project the right answer?
When the work has an end. A store build, a migration, a feed rebuild, a measurement repair: these finish, and paying monthly for something that finished is how agencies end up billing for maintenance nobody asked for. Buy the project as a project. Keep the ongoing fee for the work that is genuinely ongoing, which is media buying, email and the measurement around them.
What does a commission on growth actually mean?
The agency is paid on the increase, so it earns nothing for holding the account still. That is the cleanest alignment available, and it has one condition attached that you should insist on before agreeing to it: the growth has to be measured against a baseline agreed in writing before anything starts, on a revenue number that comes from your shop rather than from an ad platform. Without a fixed baseline, a commission model quietly rewards a good season. With one, it is the only model where the agency loses money when you do.
It is what we run alongside a fixed monthly fee, and we do not take a percentage of ad spend at all. You can see how the rest of it works on the how we work page.
What actually changes the number
- Delivery or advice. Somebody building the flows, the feed and the pages costs more than somebody reviewing them, and the two get described with the same words in a proposal.
- Channel count. Each channel added is another account to run, not another line on an invoice. A brand on Meta, Google, email and search is buying four jobs.
- Cadence. Weekly attention and monthly attention are different products. Ask how often somebody opens the account, and what happens in between.
- Who writes the report. A generated dashboard costs the agency nothing. A person explaining what they changed and why costs real hours, and it is the only part of the reporting that has ever told anybody anything.
What to ask before anything is signed
Ask three things and ask for them in writing. Which model, and what happens to the fee when your spend goes up or down. What exactly is delivered each month, in verbs rather than in channel names. And what the exit looks like: notice period, who holds the accounts, what you keep. An agency that answers those three plainly has told you more about the price than any figure would, because the figure follows from the answers.