Abstract: In view of the problem that the massive growth of marketing data of integrated energy service providers makes it difficult to effectively classify customer types and lacks corresponding value-added service measures, this paper proposes a comprehensive energy customer classification and value-added service strategy based on multi-feature labels and improved K-Means algorithm. First, based on comprehensive energy service marketing data, five feature labels including energy usage characteristics and customer value are set to comprehensively evaluate customer energy usage behavior. Secondly, CH evaluation indicators are introduced to improve the traditional K-means algorithm to achieve comprehensive energy Accurate classification of customers. Finally, based on the clustering results, differentiated value-added services are provided for different types of customers. Through experimental analysis of actual marketing data of comprehensive energy, the results show that the setting of feature labels can comprehensively evaluate customer energy usage behavior to achieve preliminary classification of customer segments when dealing with massive data. Compared with traditional clustering algorithms, the customer classification of the improved K-means algorithm is more accurate. Through targeted value-added service design, it can provide enterprises with better resource allocation and service strategies to better meet the energy needs of different customer groups.