Mobile Dev Memo Podcast Season 5, Episode 3: Marketing experiments and probabilistic measurement (with Koen Pauwels and Julian Runge)
Jan 14, 2025
In this discussion, Koen Pauwels, a marketing professor at Northeastern University, and Julian Runge, an assistant professor at Northwestern, dive into the nitty-gritty of marketing experimentation versus probabilistic measurement. They highlight the importance of randomized control trials and how firms often misapply Media Mix Modeling. The duo also explores ways to measure brand equity, emphasizing a balance between quantitative and qualitative insights. Additionally, they shed light on effectively demonstrating marketing value to finance teams and integrating new measurement techniques in a rapidly changing digital landscape.
AI Snips
Chapters
Transcript
Episode notes
Framing Marketing Measurement
- Consider the specific business question you're asking.
- Determine the most interesting and feasible counterfactual for your situation.
Seufert's View on Marketing Measurement
- Eric Seufert presented his view on experimentation and media mix modeling, focusing on reducing omitted variable bias.
- He discussed using time series for forecasting and RCTs for estimating counterfactuals.
Misinterpreting Experimental Results
- At Amazon, Koen Pauwels observed advertisers misinterpreting experimental results.
- Advertisers would halt all advertising based on one insignificant campaign test, ignoring other factors.