With the rapid development of emerging adjustable loads, flexible resources within low-voltage user groups, including commercial, industrial, and residential sectors, have become an important potential source for enhancing grid flexibility. However, the complex and heterogeneous characteristics of user-side resources make it challenging to accurately identify and quantify their adjustment potential. At the same time, traditional pricing mechanisms present inherent limitations and fail to provide effective incentives for differentiated user groups. To address these challenges, this study proposes a precision demand response mechanism for low-voltage users based on multi-dimensional feature clustering and dynamic game-based pricing. A comprehensive user profiling framework is first established, incorporating load characteristics, photovoltaic generation, and energy storage behavior. An improved K-means clustering algorithm is then employed to achieve precise user segmentation, followed by the development of a composite scoring system based on user features. In addition, a dynamic pricing model is formulated within the Stackelberg game framework, in which equilibrium prices are determined through strategic interactions between aggregators and users. The results demonstrate that the proposed mechanism can significantly enhance the participation of demand-side resources, improve the operational stability of the power grid, and ensure a balanced distribution of benefits among users, aggregators, and grid operators.