This paper evaluates the effectiveness of three feature selection techniques—Recursive Feature Elimination (RFE), Correlation Analysis, and Lasso Regularisation—in enhancing K-means clustering for B2B customer segmentation. Using a quantitative case study approach, the research assesses the individual and combined impact of these methods on clustering performance. The dataset, comprising anonymised B2B interactions from a wholesale distribution company, presented a high-dimensional and complex environment in which to test these techniques. Findings indicate that a hybrid approach—applying Lasso Regularisation, RFE, and Correlation Analysis in sequence—outperforms the individual methods. This integrated strategy improves silhouette scores and cluster cohesion, resulting in more accurate and interpretable segmentation. The study demonstrates that combining these techniques produces a robust framework that yields actionable insights for targeted marketing, resource allocation, and customer engagement within B2B contexts.