User satisfaction is a common evaluation metric in task-oriented dialogue systems, whereas tutorial dialogue systems are often evaluated in terms of student learning gain. However, user satisfaction is also important for such systems, since it may predict technology acceptance. We present a detailed satisfaction questionnaire used in evaluating the BEETLE II system (REVU-NL), and explore the underlying components of user satisfaction using factor analysis. We demonstrate interesting patterns of interaction between interpretation quality, satisfaction and the dialogue policy, highlighting the importance of more fine-grained evaluation of user satisfaction.