Suspicious online product reviews: An empirical analysis of brand and product characteristics using Amazon data

Abstract

Concerns over the authenticity of reviews hinder their usefulness. Consumers discount the information they get from reviews, and this particularly harms brands with high ratings. The authors argue that perceived brand strength, brand advertising effort, price, and sales each act as signals that help positive reviews of a brand seem more reliable. They investigate this by studying Amazon.com reviews of branded products from 16 product categories that have the resources and potential desire to advertise. They examine consumer perceptions of review authenticity as perceived by machine learning algorithms trained on human subjects, as well as by direct perceptions of human subjects during validation. The results indicate that having a strong brand and investment in brand advertising, along with price and sales rank, are valuable to managers since they form a certain protection from suspiciousness.