Highly automated test vehicles are rare today, and (independent) researchers have often limited access to them. Also, developing fully functioning system prototypes is time and effort consuming. In this paper, we present three adaptions of the Wizard of Oz technique as a means of gathering data about interactions with highly automated vehicles in early development phases. Two of them address interactions between drivers and highly automated vehicles, while the third one is adapted to address interactions between pedestrians and highly automated vehicles. The focus is on the experimental methodology adaptations and our lessons learned.