In the coming years teachers are going to use robots as learning tools in education because of their beneficial outcome in children's practical and cognitive skills even in special education. Apart from using social educational robots as a tool, teachers are expected to collaborate with them in many ways, even as a peer teacher in the learning procedure and get involved into the robots' design process. Although those procedures are progressive, in many cases, teachers cannot provide very useful feedback because they do not have prior experience with robots operating within the target environment. Therefore, it is important to identify the best way to accurately identify the teachers' needs. In this paper, we introduce the idea of actively involve the future teachers in the design process by having them watching a robot performing the target task. Our target is to find the ideal robot-tutor characteristics from the teachers' perspective in order to collaborate with it and we compare the future teachers' opinions about the ideal characteristics (Appearance, Intelligence, Emotional Expression etc.), when they are actually taught by a social robot and when they are taught by a human tutor. Participants had statistically significant different opinions about the robot's characteristics after each class, such as their preference for machinery robots after the human-class which is non-existent after the robot-class. Our results clearly suggest that teachers as robots' future users should interact with a similar robot performing the target task, in order to be able to accurately select the target robot characteristics.