Decomposing the Variance of Consumer Ratings and the Impact on Price and Demand

Consumer ratings play a decisive role in purchases by online shoppers. Although the effect of the average and the number of consumer ratings on future product pricing and demand have been studied with some conclusive results, the effects of the variance of these ratings are less well understood. We develop a model which considers durable goods that are characterized by three types of attributes: search attributes, experience attributes, and transformed attributes the latter are conventional experience attributes that are transformed by consumer ratings into attributes that can be searched. Using informed search attributes to refer to the combination of search attributes and transformed attributes, we consider two sources of variance of consumer ratings: taste differences about informed search attributes and quality differences in the form of product failure representing experience attributes. We find that (i) optimal price increases and demand decreases in variance caused by informed search attributes, (ii) optimal price and demand decrease in variance caused by experience attributes, and (iii) by holding the average rating as well as the total variance constant, for products with low total variance price and demand increase in the relative share of variance caused by informed search attributes. Counter to intuition, we demonstrate that risk averse consumers may prefer a higher priced product with a higher variance in ratings when deciding between two similar products with the same average rating. Finally, our model provides a theoretical explanation for the empirically observed j-shaped distribution of consumer ratings in e-commerce that differs from established explanations.