On Information Distortions in Online Ratings

Abstract. Consumer reviews and ratings of products and services have become ubiquitous on the Internet. This paper analyzes, given the sequential nature of reviews and the limited feedback of such past reviews, the information content they communicate to future customers. We consider a model with heterogeneous customers who buy a product of unknown quality and we focus on two different informational settings. In the first setting, customers observe the whole history of past reviews. In the second one they only observe the sample mean of past reviews. We examine under which conditions, in each setting, customers can recover the true quality of the product based on the feedback they observe. In the case of total monitoring, if consumers adopt a fully rational Bayesian updating paradigm, then they asymptotically learn the unknown quality. With access to only the sample mean of past reviews, inference becomes intricate for customers and it is not clear if, when, and how social learning can take place. We first analyze the setting when customers interpret the mean as the proxy of quality. We show that in the long run, the sample mean of reviews stabilizes and, in general, customers overestimate the underlying quality of the product.

Introduction. 1.1. Motivation The use of reviews has become ubiquitous on the Internet, where websites allow users to comment on the products or services they purchased. Typically the review consists of a grade and a written comment. The grade may be expressed in different ways, the most common one being the five star system, used, among others, by Amazon, Yelp, and Expedia. Such grades, or their aggregate statistics, are often the basis for other users’ decisions, who in turn may post their reviews. The impact of reviews on rated services or products has been reported to be significant, for example, attributing 5%–9% increase in revenue to an increase in one star on Yelp (Luca 2011). Reviews are used by consumers before making a decision to purchase an item. Once a consumer has purchased the item, she may write a review that expresses her level of satisfaction. Typically consumers do not know the exact quality of the object that they intend to purchase, for instance they either ignore or would not be able to interpret the technical features of a new smartphone.

Discussion / Conclusion. 5. Extensions and Conclusions Reports versus sincere evaluations. Various studies have documented that observed statistics may have an important impact on consumer ratings. For example, Talwar et al. (2007) show that a user’s rating partly reflects the difference between true quality and prior expectation of quality as inferred from previous reviews through some empirical analysis. Moe and Trusov (2011) show, through an analysis of sales data across time, that the current average of reports has a significant effect on consumer rating and sales. This raises the question of whether past reviews not only impact purchasing decisions (as captured in the previous sections) but also impact what a customer ultimately reports. In turn, this leads to the question of whether social learning can take place if reviews are not truthful.