Ancillary Services in Targeted Advertising: from Prediction to Prescription
38 Pages Posted: 1 Jun 2022 Last revised: 3 Jun 2022
Date Written: September 17, 2020
Abstract
Problem definition: Online retailers provide recommendations of ancillary services when a customer is making a purchase. Our goal is to predict the Net Present Value (NPV) of these services, estimate the probability of a customer subscribing to each of them depending on what services are offered to them and ultimately prescribe the optimal personalized service recommendation that maximizes the expected long-term revenue.
Academic/Practical Relevance: The market of online retail has reached $1.5 trillions in the United States alone. This is even more the case with COVID-19. In fact, the impact of personalization and display optimization cannot be overstated. Our framework not only offers a holistic, high-performance solution to this problem, with an immediate and transformative impact for our industry partner, Wayfair, but is also generalizable to a large set of problems involving estimating an objective value, predicting propensity to buy and finally prescribing an optimal decision.
Methodology: We propose a novel method called Cluster-While-Classify (CWC), which jointly groups observations into clusters (segments) and learns a distinct classification model within each of these segments to predict the sign-up propensity of services based on customer, product and session-level features. This method is competitive with the industry state-of-the-art and can be represented in a simple decision tree. This makes CWC interpretable and easily actionable. We then use Double Machine Learning (DML) and Causal Forests to estimate the NPV for each service, and finally propose an iterative optimization strategy — that is scalable and efficient — to solve the personalized ancillary service recommendation problem.
Results: CWC achieved a competitive 74% out-of-sample accuracy over 4 possible outcomes and 7 different combinations of services for the propensity predictions, which, alongside the rest of the personalized holistic optimization framework, resulted in an estimated 2.5-3.5% uplift in revenue.
Managerial Implications: The proposed solution allows online retailers in general, and Wayfair in particular, to curate their service offerings and optimize and personalize their service recommendations for the stakeholders. This results in a simplified, streamlined process and a significant long-term revenue uplift.
Keywords: display optimization, NPV estimation, classification, causal inference, predictive analytics, prescriptive analytics, online retail
Suggested Citation: Suggested Citation