Essay · About a 15 minute read

Marketing Analytics: Attribution Is Not Incrementality

Avinash Kaushik · Occam's Razor Free Intermediate

This is the piece that names the exact trap you are in: your ad platforms each claim credit for the same sales, so the numbers add up to more than reality. Kaushik walks through a plain example where 10 conversions get reported but only 3 were actually caused by the marketing, a 70 percent overstatement. It is the clearest starting point for understanding why attribution and incrementality are not the same thing.

  • Attribution just splits credit across touchpoints. It never asks whether the sale would have happened anyway, which is the only question that matters.
  • A large share of conversions credited to paid search or social would have converted organically, from brand, word of mouth, or existing demand.
  • Real incrementality needs an experiment or a proper model, not a prettier attribution dashboard.
Open on kaushik.net ↗

The question it answers

  1. How do I stop platform-reported conversions from lying to me about what's actually driving sales?

Read with it

  1. Video · Podcast style conversation, roughly 30 to 45 minutes Measuring Top of Funnel and Incrementality Testing with Zach Epstein from Haus Haus (Zach Epstein)
  2. If full incrementality testing is more than your team can run right now, blended metrics are the honest shortcut, and this... Eightx Article · About a 10 minute read ROAS vs MER vs Blended CAC: Which Metric Actually Matters Matt Putra
eChai Partner Brands The partnership →