Automated Marketing Research Using Online Customer Reviews

When shopping online, consumers often read several reviews of products they discover and many times base their decisions to purchase on the information provided by reviews. Research covered in class focused on the elicitation of ratings and reviews from consumers and ensuring they are valuable to the consumer. Reviews are not only beneficial for the consumer and there are distinct benefits that companies can extract directly from the information in the reviews.

In their study Lee and Bradlow (2011) use text-mining techniques to automate the analysis of customer reviews, forming valuable information for use in market research. Previous studies have not covered the analysis of market structure through reviews to describe the environment surrounding a business. Market structure analysis is an important part of the market research process, as many of the marketing decisions rely on information about existing the existing market; the potential substitutes and complements for the product. In order to form these market structure analyses, attributes of products are commonly mapped to represent different brands.

The study utilizes simple methodology to suit capabilities of marketers better; by combining techniques commonly used and less complex language processing. The techniques chosen do not require predefined product attributes to be tested which is how current market research commonly approaches eliciting these attributes. The authors’ rationale is to allow the methodology to be used repeatedly, so that analysis “can be done (unlike traditional methods) continuously, automatically, inexpensively, and in real time.”

The authors’ collected all digital camera reviews on the epinions.com platform between  July 2004 and 2007. By clustering attributes detected in product reviews into common attributes, the authors’ were able to compare these attributes to attributes found in expert buying guides. What they found when comparing the opinions of experts, interestingly, was that they have no consensus in what attributes of a product are seen as important. In their comparison study they found attributes from analyzed reviews to be more valuable to the respondents and discovered new attributes.

To prove the use of their methodology in forming overviews of market structure repeatedly, the authors’ ran the analysis on a parallel data set they collected from reviews in between 2005 and 2007. This showed interesting results as the changes in attributes matched the changes seen in the market in terms of company strategies and consumer tastes. When Nikon changed its marketing from promoting technical specifications to a more product benefit focused approach, the attributes used in reviews reflected the change.

Managers can use the findings of Lee and Bradlow to support marketing strategy decisions. By mining customer reviews, the company can see how its brand aligns with competing brands in consumers’ minds. Tracking how the attributes mentioned change over time can be valuable information in determining how successful campaigns have been. New segments can be found by clustering characteristics detected with semantic analysis. Spotting attributes associated with competitors’ products is valuable insight in how competition is performing.

The study showcases how big data can be used in marketing research and brings to light the great value customer review data has when finding the customers to involve in the value creation process. With the current popularity of social media analysis it would be fascinating to compare the effectiveness of analyzing reviews and social media postings.

Thomas Y. Lee, Eric T. Bradlow (2011) Automated Marketing Research Using Online Customer Reviews. Journal of Marketing Research: October 2011, Vol. 48, No. 5, pp. 881-894.