If every channel is performing, where is the waste?
Meta says its AI drove 15.7% more conversions this year. The average ad also cost 12% more. DTC brands spent a quarter more on Meta and earned about the same per dollar. When every number looks fine, nobody goes looking for the money in the wrong channel.
Meta will keep getting better at Meta. Google will keep getting better at Google. Neither of them will ever see all your channels at once. You do. That is the only place left to find money.
Reported one way, modeled another.
Across 299 DTC brands last quarter, Google brand search reported 19x. Tested against holdouts, the same spend returned 5.7x. Meta went the other way.
| Channel | Reported | Tested |
|---|---|---|
| Google Brand | 19.07x | 5.72x |
| Meta Acquisition | 1.83x | 2.07x |
One channel overstated by 70 percent, the other understated. You cannot guess which way yours are off.
We model a ROAS for each channel from your own spend and revenue, and show it next to what the platform reported. Holdout tests are how you check a model like that. If you run them, send them and they go in. If you don't, the model still ranks your channels, and it shows you where a test would be worth running.
A ranked allocation, and the evidence under it.
One page of four. Download the full report →
Run it again as new sales and spend data lands, so the split keeps up instead of waiting for the next planning cycle.
Kestrel, active apparel. $3M across 8 channels, modeled against each revenue stream separately, so retail shows up where the platforms can't see it.
Luma, mass beauty. $24M across 13 channels, one ranking grouped the way the teams are organised, so each reads its own channels in the same numbers.
Send exports. About a week. Then again as data comes in.
The decision layer between measurement and action: where does the next dollar go, and what is already committed that we would have to work around?
Half of agency AI platforms will be obsolete by 2029. The other half will own the decision layer.
The difference between adopting AI tools and building an AI-native workflow that actually fits how your agency operates.