Measurement is a governance problem, not a data problem

Every platform will happily measure itself. The question is who owns the model that decides where the next dollar goes.

Marketing teams talk about measurement like it's a data problem. Better attribution, cleaner pipes, more signals feeding the model. Those things matter. But in most businesses we look at, the real problem is simpler and considerably harder to fix: the people reporting on performance are the same people who profit from how it gets reported.

The platform reports what the platform delivered. The agency reports against the plan the agency built. The attribution model was configured by whoever set up the account, often years ago, by someone no longer at the company, and it quietly credits the channels that happen to be easiest to credit.

Then the budget follows the report, year after year, because that's what the report is for.

What That Actually Produces

In one recent audit, a legacy first-touch model was crediting programmatic display with roughly twice its real contribution to revenue. Nobody had done anything dishonest. The team trusted the number because it was the only number they had, and it had been sitting there long enough to feel like fact rather than configuration. But budget allocation had been shaped by that number for years, which means real money had been quietly misallocated for years too.

This is not rare. It's closer to the default state of measurement in businesses that haven't deliberately intervened.

Two questions are worth asking now. Who on your side, not the agency's, owns the measurement model and can explain its assumptions without looking anything up? And when a platform's self-reported numbers don't reconcile with what your own data shows, does anyone notice, or does the platform's number just win by default?

If the answer to the first is "the agency" and the second is "not sure," that's not a data gap. It's a governance gap, and it's costing real budget right now.

The Fix Is a Decision, Not a Project

Someone on the client side owns the measurement model, in writing, with its assumptions stated plainly enough that a new hire could understand them. Every major platform's self-reported numbers get reconciled against that model, not accepted at face value. Incrementality gets tested where it actually matters, including in the channels where a partner would rather it didn't. And the budget decision gets made from the reconciled view, not from whichever deck arrived most recently.

None of that requires new technology. It requires deciding who is allowed to mark their own homework, and then not letting them. Get that right and the data problems become genuinely solvable. Skip it, and no amount of better data will save you, because the incentive to shade the number never went away.