Considering human-robot-interaction as a tool for investigating situated (inter-)action, this paper investigates real-world encounters of small groups of adults and children with an autonomous museum guide robot. Focusing on Question-Answer sequences, it explores how they attempt to answer the robot’s questions and the emerging interactional visitor dynamics. Video-taped recordings and log files of the system’s speech recognition are used in a combination of fine-grained micro-analysis of interaction and Conversation Analysis with the robot’s internal perspective. This analysis reveals that adults tend to assume the role of ‘participation facilitator’ and establish the child(ren) as primary co-participant(s) for the robot. The participants collaborate to answer, which requires distinguishing between the ‘answer-as-interactional process’ and the ‘answer-as-result’, causing difficulties in the social-technical system.