You lead marketing analytics at a company in Rosario and have to decide where to put the budget. The dashboard shows that brand search and retargeting "convert" like crazy, so the temptation is to pour spend there. But those channels grab the last-click credit for sales that were already going to happen: they capture demand that some other effort created. The right question isn't "which channel touched the customer last," but "what would have happened without that spend": incrementality. Harrah's and Capital One built their edge by measuring exactly that, with controlled experiments.
The situation
Attribution splits a sale's credit among the marketing efforts that touched it. The easy method —last-click or vanity metrics— gives all the credit to the final click, which systematically overvalues demand-capture channels (brand search, retargeting): those customers were often already going to buy, and the channel was just there at the end. The right question is incrementality: what sales did the spend actually cause, versus what would have happened without it? It's measured with controlled experiments and holdout groups. Harrah's used data to segment and measure each customer's real value; Capital One made test-and-learn —measuring incrementality— its competitive edge.
Here's the trap, in two directions. Optimizing for last-click and vanity metrics feels data-driven, but it measures correlation, not cause: it overfunds the demand capture that already existed and underfunds the demand creation that actually brings new customers. And trusting instinct and not measuring is flying blind: without experiments you don't know what works and you waste budget just the same. The defensible move is to measure incrementality with controlled experiments (holdouts, test-and-learn) and allocate spend to what truly generates incremental results.