Author: Luca Turin
Supervisor: Julia Neidhardt, Co-Supervisor: Ahmadou Wagne
Abstract
Understanding user behaviour is essential for designing digital services, improving user experience, and optimising commercial outcomes. This thesis investigates how behaviour varies over time and across contexts on a major price comparison platform. Building on prior work that offered a static segmentation of sessions, it examines how these patterns evolve over time. The study addresses three questions about the extent of behavioural differences: month to month, weekday versus weekend, and shifts during the Black Friday period. We analyse one year of clickstream data from July 2024 to June 2025. We segment the data by time, apply unsupervised clustering within each segment, and align labels across segments to enable comparison. The results show a stable set of behavioural segments together with clear temporal change. Month to month differences are evident, with a shift toward deal seeking during the end of year shopping period and around Black Friday. Weekends show more exploratory behaviour than weekdays. Black Friday polarises activity, with more short check in visits and more concentrated deal focused research. These patterns indicate that user behaviour changes meaningfully with time and context.