Affordable Uplift: Supervised Randomization in Controlled Experiments
This work addresses the problem of costly data collection for businesses using uplift models in marketing, offering an incremental improvement to reduce experimentation expenses.
The paper tackles the high cost of collecting randomized experimental data for training uplift models in direct marketing by introducing supervised randomization, which integrates existing scoring models to target relevant customers while ensuring consistent treatment effect estimates; an empirical study shows this approach is cost-efficient and yields competitively performing uplift models.
Customer scoring models are the core of scalable direct marketing. Uplift models provide an estimate of the incremental benefit from a treatment that is used for operational decision-making. Training and monitoring of uplift models require experimental data. However, the collection of data under randomized treatment assignment is costly, since random targeting deviates from an established targeting policy. To increase the cost-efficiency of experimentation and facilitate frequent data collection and model training, we introduce supervised randomization. It is a novel approach that integrates existing scoring models into randomized trials to target relevant customers, while ensuring consistent estimates of treatment effects through correction for active sample selection. An empirical Monte Carlo study shows that data collection under supervised randomization is cost-efficient, while downstream uplift models perform competitively.