Precision Marketing Strategies for E-commerce Platforms Based on a Density‑Optimized K‑means Algorithm

E‑commerce websites experience saturating advertisement paybacks as diverse sets of users grow. Traditional K‑means clustering with local point density agnosticism produces inflated clusters and wasted advertising budgets. We propose a Density‑Optimized K‑means (DO‑Kmeans) that incorporates reachability‑driven weighting into the update of the centroids and hence calibrates cluster granularity to underlying consumer concentration. Implemented on two million behavioral traces of a prominent marketplace, DO‑Kmeans increases click‑through rate by 12.4% and reduces cost per acquisition by 9.7% compared to MiniBatch K‑means, Gaussian‑Mixture, and DBSCAN while maintaining near‑linear scalability. The micro‑segments generated by it support hyper‑personalized coupon distribution and dynamic banner allocation together improving overall uplift in net revenue by 8.3%. A theoretical proof also demonstrates that the proposed objective maintains the convergence guarantees when compared to classic K‑means subject to bounded ratios of densities. Strong robustness against levels of noise, seasonality oscillations, and spike events is established through extensive ablation experiments. Algorithmic simplicity combined with commercial uplift outlines a clear direction towards data‑driven expansion.