Abstract

This article develops and estimates a model of online purchase behavior using clickstream data from a branded merchandise store. The model links consumers’ purchase amounts to how they browse the website in three distinct windows—day-of, short-term, and long-term—before making a transaction. Focusing on session frequency, cart usage, and product detail views, we employ a finite mixture approach to account for unobserved heterogeneity across users. Our results reveal three distinct segments of online shoppers: low-spending task-focused buyers, moderate spending explorers, and high-spending advance planners. Each segment displays unique patterns of search depth, cart interaction, and timing. We find that day-of visits drive quick purchasing among lower spenders, whereas long-term browsing and comprehensive product
reviews play a critical role for high-value orders. These findings underscore the importance of segment-specific, time-targeted strategies for digital retailers seeking to enhance basket values and tailor website design to varying consumer needs.