Predicting customer quality in e-commerce social networks: a machine learning approach

Abstract

The digital transformation of companies is having a major impact on all business areas, especially marketing, where audiences are most volatile and loyalty is at its scarcest. Many large retail brands try to keep their client base interested by becoming partners in cashback websites. These websites are based on a specific type of affiliate marketing whereby customers access a wide range of merchants and obtain financial rewards based on their activities. Besides using this mix of traditional marketing strategies, cashback websites attract new target customers and increase existing customers’ loyalty through recommendations, using a word-of-mouth marketing strategy built on economic incentives for users who refer others to these sites. The literature shows that this strategy is one of the major areas of success of this business model because customers who join following recommendation are more active and are therefore more profitable and loyal to the brand. Nevertheless, the new users who are referred to these sites vary considerably in terms of the number of transactions they make on the site. This study advances research on the design of recommendation-based digital marketing strategies by providing companies with a predictive model. This model uses data science, including machine learning methods and big data, to personalize financial incentives for users based on the quality of the new customers they refer to the cashback website. Companies can thus optimize and maximize the return on their marketing investment.