What you are actually comparing
| Agency retainer | Platform alone | Platform plus an operator | |
|---|---|---|---|
| Execution | Yes | Yes | Yes |
| Attention between reviews | Weekly at best | Continuous | Continuous |
| Strategy | Yes | No | From the operator |
| Accountability when it goes wrong | Contractual | Yours | Yours |
| Cost shape | Scales with spend | Banded | Banded plus salary |
The middle column is the one that flatters software and misleads buyers. A platform with nobody directing it is not a cheaper agency, it is a faster account with no one deciding what it is for. The right-hand column is the real comparison, and it is the one worth doing on your own numbers.
Where the platform wins
Attention. An agency reviews your account on a cadence set by how many other clients the team has. A system checks continuously. Most money lost in an ad account is lost in the gap between reviews, on a Tuesday, quietly.
Consistency. The checks run identically every time. Nobody is on holiday, nobody has just been assigned three new accounts, nobody skipped the tedious one.
Cost shape. The fee is set by the size of the account and does not rise because you increased budget, which removes the conflict built into percentage of spend pricing.
Breadth. Reading paid, analytics, email and orders together is routine for software and expensive in people, because it means someone senior looking across four tools at once.
Where the agency wins
Below a certain size. A small account cannot justify a platform fee plus someone to run it, and does not generate enough data for continuous optimisation to beat a competent person doing a monthly review. Say so out loud: if you are spending a few thousand a month, a good freelancer is better value than any of this.
Strategy and creative. Deciding what to say, to whom, and what the brand will not do is not an optimisation problem. Software does not do it and mostly should not pretend to.
Accountability. You can shout at an agency. You can replace an agency. A platform that lost you money last month is a support ticket.
Nobody to run it. The platform assumes a person who sets the targets and answers the approvals. If that person does not exist and is not being hired, the agency is doing something you genuinely cannot cover.
The comparison most buyers get wrong
Setting the retainer against the subscription and declaring a saving ignores the part of the retainer that was buying judgement. The saving is real only if that judgement is coming from somewhere: an in-house marketer, a founder with the time, or a fractional strategist.
The honest arithmetic is retainer against subscription plus whatever fraction of a person now directs the system. Sometimes that still favours software comfortably. Sometimes it is a wash and you should stay put.
The version nobody sells you
The two are not exclusive, and the arrangement that works best is usually both: the agency keeps the strategy and the creative, and the platform takes the execution and the monitoring the agency was doing manually and charging for. That makes the retainer smaller and the work better, which is a conversation worth having with an agency you rate.
It is also what a good agency does with this: running it across client accounts rather than being replaced by it. Whether that improves their margin or their service is up to them, and the ones who treat it as the second tend to keep clients longer.
How to work it out for your account
- Take the retainer, annualised.
- Add what the agency's tools cost you separately, if anything.
- Against that, put the subscription plus the salary fraction of whoever will direct it.
- Ask what happens to each number if your spend doubles.
- Ask what happens to each if the person running it leaves.
Step four is where percentage-of-spend arrangements diverge sharply, and step five is where the platform usually wins: the pricing model tells you more about the next three years than the current figure does.