The $2M Mistake

We see this pattern regularly when reviewing test results with scaling brands. They run a multi-cell lift study comparing two strategies. The lift results identify a winner. Then they pull up the same test period in their attribution dashboard, and it tells them the opposite.

Take two strategies, tested head-to-head in a multi-cell lift study over four weeks:

Attribution picks A. The lift study confirms B. That's not a rounding error. That's a fundamentally different answer to "where should I put my next dollar?"

A $200K gap per test period. Let that gap run for a year and you're looking at over $2M in missed incremental revenue.

The reason this happens is predictable. Even modeled attribution, which goes beyond clicks to include view-through and statistical estimates, still assigns credit based on correlation, not causation. The gap between modeled attribution and incrementality is typically smaller than with pure last-click, but it's still real.

Retargeting often looks better in attribution models because those audiences are already close to converting. Prospecting often looks weaker for the opposite reason. Video often underperforms. Static often over-indexes. The measurement system reshapes the strategy, and the strategy reinforces the measurement system.

That's the loop you need to break.

The calibration multiplier (incremental results divided by attributed results) from a lift study doesn't just adjust a number. It reveals which strategies are driving growth and which ones are performing for your dashboard but not for your business. A calibration multiplier applied at the tactic level corrects magnitude. But to catch winner reversals like the one above, you need head-to-head multi-cell tests that isolate which strategy actually drives more incremental growth.

If you've never compared your lift results against your attributed results side by side, start there. The gap will tell you more about your measurement health than any single metric on your dashboard.