The determination of user needs and their degree of importance is a critical link for designers to carry out product development, and it also plays a crucial role in guiding the success of enterprise product development. However, the current elderly care service robots only consider their directness when designing and developing. The needs of users of the elderly lack a comprehensive consideration of the needs of their children, that is, the needs of indirect users for service robots. Therefore, an alternative design demand identification method for the elderly care service robots that integrates multi-agent users is proposed to help designers clarify the direction of product replacement in order to meet the actual needs of multi-agent users, thereby enhancing the competitiveness of enterprise products. First, this paper aims to obtain the common needs of the direct users and the children of indirect users through user interviews, namely the key needs. Second, this papre uses the entropy method to determine the initial importance of the key needs, then uses the fuzzy kano model to determine the key needs category, and finally classifies the results according to the key needs, the importance adjustment function is introduced to adjust the initial importance of key needs, and the importance of the needs of the elderly and their children is comprehensively considered, and the needs that do not meet the expectations of the users are identified, so as to discover the deficiencies of the current the elderly care service robots, and guide designers to carry out iterative design to help companies develop the elderly care service robots that meet the actual needs of multi-agent users. In conclusion, through a comprehensive analysis of the multi-party needs of the elderly care service robot, it has clarified its shortcomings in medication guidance, remote doctors, and audio-visual entertainment, helping designers find the direction of iterative design, and verifying the effectiveness and operability of the method. Furthermore, this method also provides a certain reference value for other improved designs involving multi-agent user products.