Is Products Recommendation Good? An Experiment on User Satisfaction

Recommendation systems may use different algorithms to present relevant information to users. In e-commerce contexts, these systems are essential to provide users with a customized experience. Several studies have evaluated different recommendation algorithms against their accuracy, but only a few evaluate algorithms from the user satisfaction viewpoint. We here present a study that aims to identify how different recommendation algorithms trigger different perceptions of satisfaction on users. Our research approach was an experiment using products and sales data from a real small retailer. Users expressed their satisfaction perception for three different algorithms. The study results show, that the algorithms proposed did not trigger different perceptions of satisfaction on users, giving clues of improvements to small retailers websites.