How Marketing Attribution Misses Incrementality on Amazon

Episode notes

Lucas and Luna dive into why standard marketing attribution models are blind to Amazon's ecosystem. They explore the 'last-click Amazon problem,' where brands see a surge in sales after ads but can't prove causality. Using the example of a D2C skincare brand that ran a six-week Amazon campaign, they break down how Amazon's attribution tools overcredit its own ads while ignoring organic search and external drivers. Key takeaways: Amazon's 'attribution window' is too short (only 14 days), and its 'new-to-brand' metric excludes existing customers. The episode explains how brands are using incrementality testing—geo-lift studies and pseudo-experimental designs—to measure true Amazon ad effectiveness. Specific data point: a recent study showed Amazon ads deliver only 1.2x incremental return, far below the 4x reported by Amazon's dashboard. Lucas and Luna argue that without proper incrementality measurement, marketers overinvest in Amazon and underinvest in other channels.