Competing for Amazon’s Buy Box: A Machine-Learning Approach

Abstract

A key feature of the Amazon marketplace is that multiple sellers can sell the same product. In such cases, Amazon recommends one of the sellers to customers in the so-called ‘buy-box’. In this study, the dynamics among sellers for occupying the buy-box was modelled using a classification approach. Italy’s Amazon webpage was crawled during ten months and features from products analyzed to estimate the more relevant ones Amazon could consider for a seller occupy the buy-box. Predictive models showed that the more relevant features are the ratio between consecutive prices in products and their number of assessment received by customers.