Classifying Customers’ Journey from Online Reviews of Amazon Fresh via Sentiment Analysis and Topic Modelling

Abstract

A positive customer journey experience is necessary to maintain customer loyalty in online retailing. After the outbreak of Covid-19, there has been a significant increase in the number of customers who buy online groceries. Due to the anonymity and convenience throughout the customer journey, E-grocery shopping platforms have become a reliable source for gathering online customer reviews. In the study, we used text mining and machine learning (ML) models to an e-grocery customer review database from the Amazon Fresh website to forecast customer feelings in the data set. To be more specific, this study aimed to determine whether the customers are satisfied with the online purchase of products or not. Further, the study aims to analyze whether the customers would recommend the purchased products or not. For sentiment analysis a sample of 78,619 reviews was used. We used a linguistic approach consisting of ML and dictionary scoring algorithms to forecast customers’ sentiment based on their reviews. Topic modeling (TM) on 3,26,120 customer reviews was used to reveal “themes” from customer reviews to grasp a better knowledge of customers experiences.