The incentive problem nobody in this industry wants to discuss
An agency paid a percentage of your media spend has an arithmetic reason not to find out that some of it is unnecessary. This is not cynicism; it is structure.
iStudios, Strategy practice
Most media agencies are remunerated, in whole or in part, as a percentage of what their clients spend. It is a long-standing model with a simple consequence: the recommendation that reduces spend also reduces the agency's revenue.
This does not require anyone to behave badly. It requires only that, faced with two defensible readings of ambiguous data, an organisation tends towards the one that is better for it. Everybody does this. That is why the structure matters more than the character of the people inside it.
What it looks like in practice
- Incrementality testing is proposed, scoped, and never quite prioritised, because a test can only reduce the answer.
- Branded search sits inside the blended return figure, where it flatters everything around it.
- Channel expansion is always the recommendation, and consolidation never is.
- The annual plan starts from last year's split plus a growth assumption, which is the one methodology guaranteed not to find waste.
The test of an agency is not whether it has ever told a client to spend less. It is whether it has ever told a client to spend less with them.
The alternatives, honestly assessed
Fixed fees remove the distortion but introduce another: an agency on a flat fee has no direct reason to grow your business. Performance fees tie payment to an outcome, which sounds ideal and in practice creates an argument about attribution that neither side can win.
We work on retained and project fees, and we think it is the least bad option rather than a perfect one. The important thing is not which model you choose but that you know which way it bends, and that you ask your agency to tell you.
If they cannot describe their own incentive problem, they have not thought about it.