The concept of dividing point between configuration and variant design was proposed firstly. An optimization objective function was constructed by combining product performance and the customer satisfaction. A judging method was designed with a performance index for checking the dividing point and performance similarity for finding the parts of required variant design. Then, with the objectives of optimal product performance, a multi-objective optimization algorithm was designed based on genetic algorithm, and the process of checking the dividing point was also specified. In the end, the proposed genetic algorithm was verified by an example. The results showed that the proposed approach could solve this problem efficiency. 