University of Qom Journal of Data Analytics and Intelligent Decision-making 3115-8161 2 3 2026 07 01 Design of an Intelligent System for Optimal Discount Allocation in E-commerce Based on Customer Purchase Behavior 142 153 4829 10.22091/jdaid.2026.16144.1071 EN Ali Sarabadani PhD in IT Engineering, Department of Information Technology Engineering (E-commerce) ,Qom university, Qom, Iran. 0000-0002-3521-1401 Mahtab Tajvidian Department of Information Technology Engineering (E-commerce) ,Qom university, Qom, Iran. Fatemeh Sadeghi Department of Statistics, Faculty of Statistics, Mathematics and Computer Science, Allameh Tabataba&#039;i University, Tehran, Iran. Mehrdad Saffarie Department of Information Technology Engineering (E-commerce), K.N. Toosi University of Technology, Tehran, Iran. Journal Article 2026 06 06 This study presents an intelligent, data-driven framework to optimize discount allocation strategies within e-commerce environments. Rather than deploying standard uniform promotional tactics, this research proposes a prescriptive model that segments customers based on transaction history to maximize aggregate profitability and mitigate high-value customer attrition.Utilizing an empirical transaction dataset containing 4,570 validated purchase records from 1,000 unique customers, we calculate Recency, Frequency, and Monetary (RFM) vectors. Following min-max normalization, K-means clustering partitions the customer base. The optimal cluster structure (K=4) is formally validated using the Silhouette Coefficient (S) and the Davies-Bouldin Index (DB) alongside the traditional elbow method. We then formulate a constrained non-linear optimization model utilizing a logistic customer response function to analytically derive segment-specific discount rates that maximize expected net profit under a fixed marketing budget constraint. The empirical system identified four distinct customer archetypes: loyal high-value, at-risk, recent low-frequency, and inactive low-value customers. Compared to a baseline uniform 10 percent discount policy, the proposed optimized framework increased total profit from $184,965.74 to $197,556.92, representing an 8.7 percent improvement. Net profit margins expanded from 21.8 percent to 23.4 percent.Concurrently, high-value customer churn decreased by 12.3 percentage points, a reduction proven statistically significant via McNemar's test (p < 0.001), while promotional waste in low-value segments was entirely eliminated.These findings demonstrate that integrating behavioral clustering with constrained optimization structures yields significantly higher economic returns and customer retention compared to uniform pricing approaches. The proposed system is highly interpretable, computationally efficient, and directly applicable for deployment within small-to-medium enterprise (SME) e-commerce platforms. https://jdaid.qom.ac.ir/article_4829_3eecdc6152799bc2d1b5a84ae4095733.pdf