From clicking to consideration: A business intelligence approach to estimating consumers' consideration probabilities

Abstract

With rapid advances in e-commerce applications and technologies, finding the chance that a product falls into a consumer's consideration set after being inspected (i.e., consideration probability, CP) becomes an important issue of recommendation services and marketing strategies for both academia and practitioners. This paper proposes a novel business intelligence (BI) approach (namely, the two-step estimation approach, TEA) to estimating CPs with a two-step procedure: one is to introduce partial belongings of consumers to the latent classes with both positive and negative preferences (tastes); the other step is to generate CPs based on the degrees of partial belongings in a weighted probability manner. Experiment results from different online shopping scenarios reveal that TEA is effective and outperforms the traditional latent class model.