A Simple Model with Seasonal Effects for Noncontractual Settings

One problem with the classic models of repeat buying in noncontractual settings is that they fail to account for seasonality in the flow of transactions – a critical problem that can lead to poor forecasts and misleading managerial inferences. Several recent papers have addressed this issue but all are hampered by substantial computational burdens that greatly limit their applicability for full-scale commercial applications. We examine and exploit a nested variant of one of these models, i.e., the “seasonal model with dropout” (SMDO) that performs almost as well as Wünderlich et al.’s (2022) full HSMDO model. Its simplicity facilitates a valuable benefit, i.e., simple closed-form expressions for the key quantities of interest for any analyst working with such models. This means we can use standard maximum likelihood methods for parameter estimation and generate critical managerial inferences in near real time. We show how the model parameters can be estimated using simple cohort-level summaries of buying behavior, thereby further reducing the cost of implementation, and enabling the use of a model that allows for seasonal variation in the flow of transactions to a much broader audience.