Predicting Online Books Sales from Review Sentiments

The past decades have seen the advent of online book retail and its explosive growth. As the popularity of online book retail continues to increase, so does the interest in understanding the predictors of online book sales. Accordingly, prior research has found evidence for the significant effects of traditional (I.e. expert reviews) and social earned media (i.e. online customer reviews) on online book sales. By literature review, the researcher pinpoints three main drawbacks of the extant literature. First, scholars have only been studying the effect of online customer reviews and expert reviews in isolation, regardless of the fact that they actually interact with each other while influencing online book sales in reality. Second, little research has been carried out to closely examine the impact of sentiment reviews on online book sales with sophisticated automated sentiment analysis technique. Third, the moderating role of book category was insofar neglected in scholarship. Drawing from these perspectives, this research sets out to investigate the differential impacts of traditional earned media (i.e. expert reviews) and social earned media (i.e. online customer reviews) on Amazon.com’s book sales. Additionally, it seeks to understand the moderating effect of book category on the impact of the positive sentiment in online customer reviews and expert reviews on Amazon.com’s book sales. The research is guided by the following research question: To what extent do traditional (i.e. expert reviews) and social earned media (i.e. online customer reviews) predict Amazon.com’s book sales? To answer the research question, the researcher conducts a quantitative research which consists of two steps of analyses. First, she performs a sentiment analysis on 1,133 reviews, using the sentiment analysis software LIWC 2015 (Linguistic Inquiry and Word Count), in order to detect the percentage of positive and negative sentiment in their textual content. Next, she conducts regression tests and z-score tests for regression coefficients to assess the hypotheses. The research provides practical recommendations for online book retailers to tailor their online communication strategy and budget so as to enhance sales performance. It, likewise, produces some important theoretical implications for future research to investigate the interplay between online customer reviews and expert reviews in influencing online book sales.