Sequential Optimization for Prospective Customer Segmentation and Content Targeting

Abstract

ResMed is a global leader in medical devices for the treatment of obstructive sleep apnea (OSA). Due to the high prevalence and underdiagnosis of OSA, a key pillar of ResMed's business strategy is to increase awareness of the disease and encourage treatment. This work seeks to optimize an emerging OSA awareness channel for ResMed: online paid advertising. Specifically, a sequential optimization approach (batched sequential model-based algorithm configuration, or B-SMAC) is developed to automatically and intelligently target online advertisements through iterative batch experimentation. The result, verified through simulation and field experiment, is the maximization and characterization of ad performance over a search space of 960 mutually exclusive customer segments. Further, re-aggregation methods are developed and tested in order to transform the outputs of B-SMAC into an economically viable targeting strategy for an online ad platform, leading to improved ad effectiveness when compared to baseline strategies. These results are a proof-of-concept for sequential optimization-based ad targeting and represent a promising future direction for increasing the number of patients entering ResMed's diagnostic funnel and receiving life-altering OSA treatment.