With the gradual advancement of the digital transformation of power grid, the power marketing service mode has been continuously improved. In order to solve the problems of single service mode and poor user experience in power marketing service, a power user service method based on cohesion hierarchical cluster ing is proposed. Firstly, the power data and behavior data of powerusers are cross analyzed to obtain the typical data characteristics of power users. Secondly, the agglomerative hierarchical clustering algorithm is used to analyze the power users, evaluate the preference and power consumption trend of power users, and draw a picture of power users. On this basis, different marketing service strategies are formulated for different types of power users to improve the satisfaction of power users. Finally, when the method is applied in a city,the online penetration rate of power users is increased from 38.2%to 69.1%, and the user satisfaction is increased by 41.5%. The results verify the effectiveness of the power user service method.