Integrated design of product families has become an essential concern for successful product design and development under the trends of mass-customization and business competition. A product family means here a series of models that are offered to a variety of customers' needs with sharing some common modules. Such commonalization enables the economy of scales even under customization for individual customers, but too much commonalization may destruct flexibility for offering various products. This kind of design problems is represented with discrete conditions on module combinations and continuous conditions on module contents and multi-objective under the above tradeoffs. Further the decision must be made in the conceptual stage without detail information. For supporting such decision, this paper proposes an optimal design exploration approach, which is composed of a multi-objective genetic algorithm for mixed integer programming problems, a data mining scheme with self-organizing maps and interpretation of representative modes of design alternatives into design guidelines. The proposed approach is applied to a simplified design problem of universal motors for demonstrating its promises and effectiveness.