Quantifying the influence of consumer reviews via text analysis

Abstract

Proliferation in the e-commerce market has made consumer reviews a valuable source of data for business. Indeed, reviews have become a goldmine of information reflecting customer sentiment and can influence buying decisions. Quantification of the influencing capacity of review text would help to categorise and identify influencing centres in textual data. This paper attempts this quantification in two stages: (1) an exploration of the aspects, beyond the antecedent of review helpfulness, that inform the persuasiveness of review text; and (2) text-mining these aspects from the text corpus to compute a review influencing capacity score (RICS). An explorative survey on smartphone reviews indicates pertinent aspects to be reference to the brand and its competitors; reference to providers of online e-commerce platforms; and discussions of price with respect to product features and/or brand. Quantification is achieved using a hybrid of analytical hierarchy process and text-mining with Microsoft VBA. The score’s polarity is determined via multiplication by overall review sentiment.