When you Said 'Attribution', did you Mean 'Contribution'?
summary
INCRMNTAL presents a method that measures incrementality without user-level data and shows budget allocation must target contribution not raw attribution. It demos using daily optimization changes as mini experiments, exposes a real example where two channels each report 10,000 conversions but combined deliver no net lift (cannibalization), and uses marginal ROAS to find diminishing returns. It recommends combining attribution, incrementality and MMM and releases a white paper to back the approach.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:17Attribution no longer works
- 8:36Budget on contribution
- 10:23Measuring contribution
- 11:25Experiments for budget
- 12:29Marginal costs and returns
- 14:40Attribution versus contribution
- 15:41Summary
related talks
22mMeasurement in a Post Privacy Era
18mSession: The Science Of Success – Driving Growth With Prediction, Incrementality, And Scalable....
30mMeasurement in a Privacy First World
40mNext Best Action: Maximizing Customer Value
34mA Human Touch to Consumer Experience Powered by Data
34mCommunity Data: What to Measure and Why?
47mAccounting for user behavior data in A/B testing metrics
28mMeasurement and Analysis for Sustainable Service Delivery