Offline Data-Driven Recommender Systems for Improving Small Business Marketing Strategies

Recommender systems play a crucial role in enhancing user engagement across domains such as e-commerce, social media, entertainment, and education. Recently, they have also been used in marketing to identify high-value customers and personalize campaigns. However, small businesses often struggle with the high cost per action (CPA) and low conversion rates (CR) associated with online marketing platforms. To address this challenge, we propose a novel recommender system that leverages offline interaction data to identify customers likely to use discount coupons, thereby increasing CRs and reducing marketing costs. We address technical challenges such as cold-start problems and data sparsity by introducing tailored data augmentation techniques. The effectiveness of our approach is validated through experiments using store-level coupon and point log data, evaluated with metrics including CPA, CR, and root mean squared error. Results show that our system significantly outperforms conventional online marketing platforms, emphasizing the value of incorporating offline data with proper augmentation for cost-effective marketing.