Platforms undercount, overclaim, and double-count each other. What to do about attribution without buying another dashboard.
A familiar meeting: Meta reports a 4x return, Google reports a 3x return, and total revenue is a long way short of what those two numbers imply. Nobody is lying. The reporting is just answering a different question to the one being asked.
Two distortions run in opposite directions. Browser-side pixels undercount, because privacy changes and tracking prevention mean a meaningful share of conversions never get reported back at all. Meanwhile every platform claims credit generously within its own attribution window, so the same purchase is counted by both, and some of those purchases would have happened without either.
The undercounting has a technical fix. Server-side conversion tracking sends the event from your server rather than the visitor's browser, which survives most of what breaks the pixel, and it is what makes the platform's optimisation work properly again. That last part matters more than the reporting: an algorithm optimising on partial data finds you partial customers.
The overclaiming does not have a technical fix, because it is a measurement question rather than a plumbing one. The practical answer is a blended view: total marketing spend against total revenue for the period, compared against what happens when a channel is turned down. Incrementality is a harder question than attribution, and it is the one worth asking, because the goal is not to allocate credit but to know what would stop if the spend stopped.
None of this requires another dashboard subscription. It requires deciding which number the business is actually run on, wiring the tracking to feed the platforms properly, and treating in-platform ROAS as a steering input rather than a result. When we take over an account, this is usually the first month of work, and it frequently changes the answer to what is working before a single campaign is touched.