competitive spillover elasticities of electronic word of mouth: an application to the soft drink industry

Electronic word of mouth (ewom), especially on online platforms such as twitter, is a topic of interest for many c-suite executives. yet little is understood about competitive spillover effects in ewom, especially among mature brands in fast-moving consumer goods (fmcg) markets. in this article we analyze the entire corpus of tweets of two main fmcg brands (pepsi and coke) and use dynamic factorial analysis to classify ewom into topic categories in an unsupervised manner. we then analyze how these topics influence sales, taking into account traditional marketing mix elements and endogeneity concerns. our results show that looking at ewom in an aggregate manner (positive vs. negative valence) can be misleading and mask important effects. we see strong evidence for ewom competitor spillover, depending on ewom content diagnosticity (high vs. low). we also show the presence of asymmetric ewom spillover effects depending on the typicality and directionality of brand associations.