A Recommender System Based on Omni-Channel Customer Data

Recommender systems aim to provide personalized suggestions to customers which products to buy or services to consume. They can help to increase sales by helping customers discover new and relevant products. Traditionally, recommender systems use the purchase history of a customer, e.g., the purchased quantity or properties of the items. While this allows to build personalized recommendations, it is a very limited view of the problem. Nowadays, extensive information about customers and their personal preferences is available which goes far beyond their purchase behaviour. For example, customers reveal their preferences in social media, by their browsing habits and online search behaviour or their interest in specific newsletters. In this paper, we investigate how information from different sources and channels can be collected and incorporated into the recommendation process. We demonstrate this, based on a real-life case study of a retailer with several million transactions. We discuss how to employ a recommender system in this scenario, evaluate various recommendation strategies and describe how to incorporate information from different sources and channels, both internal and external. Our results show that the recommendations can be better tailored to the personal preferences of customers.