The rise in digital transactions has driven a massive influx of data in the retail industry, making customer segmentation crucial for personalized marketing. Traditional segmentation methods often fail to capture the complex interactions between customers. This research addresses these limitations by combining machine learning with graph analytics to deliver deeper insights into customer behaviour. Specifically, RFM-based K-means clustering was employed for segmentation, and a recommendation system was implemented for targeted marketing. Additionally, graph analytics techniques, including community detection and centrality measures, were introduced to identify key influencers within the customer network. The results, evaluated using Silhouette, Davies-Bouldin, and Caliñski-Harabasz indices, demonstrated superior performance compared to traditional methods, particularly in enhancing customer insights. This dual approach improves segmentation accuracy and supports better business decisions in resource allocation and marketing, especially within the insurance sector.