Research indicates that in order for customers to properly utilize recommendation assistants (RA), they must trust the RA and enjoy the interface. However, methods for increasing user enjoyment of RAs are yet unknown. This study investigated the influences of utilitarian and hedonic factors on intention to adopt RA suggestions and the antecedents of outcome similarity and web atmospherics. User motives and attitudes towards RAs could influence their interest in features recommended by RAs. Contrary to common assumptions, customers may make unplanned purchases, rather than rational purchase. In accordance with information cascade theory, this study investigated relationships of unplanned vs. planned purchase decisions. A 2×2×2 factorial design revealed three main findings. First, information diagnosticity and enjoyment positively affected intention to adopt RA suggestions. Diagnosticity was determined by outcome similarity, and enjoyment was determined by both outcome similarity and atmospherics. Second, involvement moderated the association between adoption intention, diagnosticity, and enjoyment. Highly involved users considered enjoyment and diagnosticity when forming adoption intentions, while users with low involvement only considered enjoyment. Third, information cascades altered the relationship between adoption intention and unplanned purchases such that information cascades negated a positive relationship between adoption intention and unplanned purchases. Theoretical and managerial implications are proposed.