Latent Stratification for Advertising Experiments



Advertising incrementality experiments often suffer from noisy responses making precise estimation of the average treatment effect (ATE) and evaluation of ROI difficult. We develop a new estimator of the ATE that improves precision by estimating separate treatment effects for three latent strata -- customers who buy regardless of ad exposure, those who buy only if exposed to ads and those who do not buy regardless. The overall ATE computed by averaging the strata estimates has lower sampling variance than the widely-used difference-in-means ATE estimator. The variance is most reduced when the three strata have substantially different ATEs and are relatively equal in size. Estimating the latent stratified ATE for 5 catalog mailing experiments shows a large reduction in the variance of the estimate. We also show that customers who have made a purchase recently and have been responsive to similar advertising in the past are less likely to be in the "do not buy regardless" stratum.

Zoom link: