Estimating Aggregate Consumer Preferences from Online Product Reviews

Today, consumer reviews are available on the Internet for a large number of product categories. The pros and cons expressed in this way uncover individually perceived strengths and weaknesses of the respective products, whereas the usually assigned product ratings represent their overall valuation. The key question at this point is how to turn the available plentitude of individual consumer opinions into aggregate consumer preferences, which can be used, for example, in product development or improvement processes.To solve this problem, an econometric framework is presented that can be applied to the mentioned type of data after having prepared it using natural language processing techniques. The suggested methodology enables the estimation of parameters, which allow inferences on the relative effect of product attributes and brand names on the overall evaluation of the products. Specifically, we discuss options for taking opinion heterogeneity into account. Both the practicability and the benefits of the suggested approach are demonstrated using product review data from the mobile phone market. This paper demonstrates that the review-based results compare very favorably with consumer preferences obtained through conjoint analysis techniques.