Due to the diversification of consumer behavior, personalization of consumer demands, rapid market changes, and intense competition, precision marketing has become crucial in the fast-moving consumer goods (FMCG) industry. However, accurately identifying customers is challenging, and excessive customer segmentation can lead to high costs. Thus, we propose a more effective clustering method. Using FMCG consumption data, we combine the Self-Organizing Map (SOM) and K-means algorithms in a two-stage clustering model. The core idea of the algorithm is to use the clustering results of the initial SOM as the starting values for K-means, applied to the RFM (Recency, Frequency, Monetary) customer value model and validated using the silhouette coefficient method. The results show that the combined algorithm significantly improves the silhouette coefficient compared to the original K-means and SOM algorithms. This enhanced clustering effectiveness provides robust support for extracting valuable information from vast data and achieving precise customer segmentation in the FMCG sector, thereby enhancing marketing strategies and operational efficiency.