LORE: A large-scale offer recommendation engine through the lens of an online subscription service

The majority of online market platforms and streaming platforms such as Amazon, Netflix, Spotify,etc. offer subscription based membership plans to access some/all of their products. In order to appeal to diverse customer groups, these services typically offer more than one type of plan. In this paper, we propose solutions to optimally recommend subscription plans to maximize user acquisition constrained by user eligibility and plan capacity (limited headcount per plan) simultaneously. We achieve this through a plan recommendation model based on Min-CostFlow network optimization, which enables us to satisfy the constraints within the optimization itself and solve it in polynomial time. We present three approaches that can be used in various settings:a single period solution, sequential time period offering, and a clustering for large scale setting. We evaluate these approaches using offline policy evaluation methods and demonstrate their value. We also discuss some practical issues in the implementation and online performance.