International Journal of Scientific Research and Engineering Development

The rapid growth of big data and deep learning technologies has transformed their applications across various domains, including marketing within business administration. One critical aspect of marketing is customer churn management, which directly impacts business performance. This study presents a method to minimize customer churn and improve purchase conversion rates by distributing personalized discount coupons to customers identified with a high likelihood of churn, using real-time big data analysis. The approach begins with two-dimensional customer segmentation through cluster analysis, followed by the development of a real-time churn prediction model utilizing clickstream data for each segment. Based on these predictions, tailored coupons are generated and delivered to target customers. The effectiveness of the method was validated by measuring conversion rates and sales growth. The findings indicate that a hybrid model, integrating churn estimation with a recommendation system, outperforms traditional standalone models in predicting customer behaviour and enhancing engagement. This approach enables online retailers to boost sales and retain customers by automatically estimating churn probability and purchasing tendencies using Support Vector Machine, all while minimizing marketing costs.