The aim of this study was to examine how social or temporal-self comparison feedback, delivered in real-time in a web-based training environment, could influence the academic performance of students in a statistics examination. First-year psychology students were given the opportunity to train for a statistics examination during a semester by doing online exercises in a web-based training environment. Once connected, students received in real-time either social comparison feedback (their score was compared to the mean score of all first-year students) or temporal-self comparison feedback (their score was compared, week by week, with their own previous score). The fact that students were free to connect to the web-based training environment heightened self-regulation differences such as academic procrastination, which was considered as a moderating variable in this study. Because academic performance was measured, the students' background in mathematics and statistics was also controlled. Irrespective of the students' background, the results reveal a positive influence of social comparison feedback on statistics exam performance, but only among students who did not delay doing exercises in the web-based training environment. By contrast, temporal-self comparison feedback did not have any effect on performance. Some recommendations for optimizing the efficacy of web-based training environments can be proposed, taking into account both social comparison feedback and academic procrastination.