Group composition presents a compelling, significant, and timely topic for educators, given the widespread use of group assignments in today's pedagogical models. This paper adopts a Fuzzy Set Qualitative Comparative Analysis (fsQCA) as a systematic approach to investigate the conditions for high learning performance in marketing simulations. We combine the objective measures available in a simulation with self-reported undergraduate students' assessment of Felder-Soloman's Index of Learning Styles (ILS). The results indicate that groups formed by students with similar learning styles perform better in the marketing simulation. On the individual level, the online simulation benefits primarily visual and intuitive learners who tend to achieve higher results. Thus, we propose an educational approach to ensure that students benefit from marketing simulations regardless of their learning style preferences. The study expands the understanding of learning styles effect in group work. Marketing educators could benefit from the results while introducing simulation to first-level marketing classes.