If retail media teams trust their measurement, why do so many catch AI wasting money?
Ask a retail media team whether they can tell if their spend is working and you'll almost always get a yes: when we surveyed 500 people across retail media, 91% said they have a real way of measuring performance.
Then we asked whether they'd ever caught AI spending money it couldn't justify. 45% said yes.
Those don't really square. If nearly everybody can see what's going on, how are so many finding the problem after the money's already spent?
Nobody here is short on data
You could assume these teams just don't have the data to catch bad AI decisions. Their own answers say they have measurement in place and trust it. They're still getting burned.
Our read is that the measurement people trust wasn't built for the mistakes AI makes. Most retail media reporting tells you what happened after an ad ran, and it's a lot shakier on what the ad caused. Human buyers cover for some of that with judgment and a bit of suspicion. An agent optimizing hard against the reported number has neither, so it finds every soft spot, and the dashboard can look fine the entire time.
Buyers are moving faster than sellers
You'd probably guess the push for more AI control is coming from vendors. They own the platforms and they have automation to sell, but it's the other way around. 35% of brands and agencies would let AI raise total spend with nobody signing off, compared to 25% of sellers. The people writing the checks are closer to handing over the keys than the people selling the inventory. And they're getting there on measurement that's already let bad spend through.
What to do with this
No one is slowing down on AI. The budget's moving whether the measurement is ready or not. The issue is that people's confidence in their numbers has gotten ahead of what those numbers can actually catch, and the more decisions go to the machine, the more that costs.
One question to bring to your own team: when AI moved budget last quarter, could you say which of those moves actually made you money?
Get the full report
The full report goes further, including:
How long teams let unjustified AI spend run before anyone stepped in.
How the most autonomous buyers treat total spend compared to everyone else.
Which side of the industry is most likely to say they'll never hand AI full control.
What buyers and sellers both want AI optimizing toward.
