Talk bursts: The role of spikes in prerelease word-of-mouth dynamics

Before their launch, many new products generate word of mouth (WOM) on social media. Such WOM typically increases toward the release date and contains sudden spikes. These spikes capture manifestations of peak consumer attention and are therefore of managerial importance, yet they have not received research attention. This article is the first to provide a comprehensive descriptive treatment of WOM spikes. The authors propose a conceptual framework to present spikes as a standalone WOM dimension and explain their emergence. They employ a robust filtering procedure to detect spikes and apply it in a data set of 90,000 prerelease online WOM messages on 157 Hollywood movies. The results indicate that prerelease spikes are widely prevalent: While some of them are event-driven, emerging in response to firm-created communications (e.g., trailer release), they are far more likely to emerge spontaneously. Content analysis reveals that WOM in spikes is more positive in sentiment and is more likely to deal with factual details than is WOM outside spikes. Prerelease WOM spikes also contribute significantly to the predictability of future product sales.

Keywords bursts, content analysis, prerelease word of mouth, user-generated content, robust time series analysis