In this paper, we investigate the effectiveness of a persuasive social robot in the field. The service robot drives around a public space and offers water to people using a persuasive message. The persuasive utterances used evoke either scientific expertise (e.g. “Research shows that it is important to drink enough water during the day”) or a reference to other people’s choices (“Most people/men/women actually do take something to drink”), hence exploring the principle of social proof. Our study makes three contributions: First, we show how persuasive utterances that are successful in the lab are not necessarily persuasive in the field. Second, we show that context factors influence the effectiveness of a persuasive message, as well as the sequential placement of the persuasive message. Lastly, the extent to which people construe the human-robot interaction situation as social influences the effectiveness of the robot as a persuasive technology in general.