Smoking cessation has become a real social challenge in healthcare domain and persuasive technologies combined with ambient intelligence figure as a possible approaches against this addiction. Choosing an effective persuasive design relies on different models and principles coming from several scientific contexts. We evaluate the triggers defined by the B.J.Fogg's model in different design strategies implementing three experiments: the first using smoking tools, the second with an Android application and a third one using the same application coupled with an Arduino-based ambient device. The CRegrette system's proof of concept suggests that statistics combined with an ambient device are more effective than notifications, mirroring and self-monitoring approaches. This article furthermore proposes some design strategies and research perspectives to support further research in the field of persuasion and ambient device design.