Evaluation of a Multivariant Interface Implementation

Multivariate user interfaces offer a dynamic way to personalize users’ experiences by adapting to their preferences and behavior. Nonetheless, implementing them demands robust justifications for such an investment. Given that web shops infrequently provide various interface versions and the business implications of this approach remain insufficiently explored, testing the proposed solution in practice is essential. The verification process must cover all aspects of the mechanism, with special emphasis on choosing the most suitable clustering method and assessing the effects of UI variants on the performance indicators employed in e-commerce. On the one hand, the research should confirm the viability of the proposed concept and methods of preparing and serving customized UI variations. On the other hand, it should serve as a foundation for future research and development aimed at personalizing the user interface in e-commerce. This chapter presents the background of the research and describes the study work carried out. The first set of experimental studies concerned the selection of the optimal clustering method. The results obtained allowed the selection of two algorithms, which were applied during the second stage of the research. This included a practical verification of the impact of dedicated user interface variants on different customer groups and confirmed that the use of layout personalization in e-commerce can yield positive business results.