Going viral or falling down? Incorporating temporal dynamics into the analysis of social media posts

Marketing interest in understanding why and how different contents achieve virality has steadily increased over the last years, and scholars are moving beyond engagement focusing on which textual characteristics are able to sustain attention. Unfortunately, analytical models generally measure the effects “ex-post”, leaving untouched the ongoing dynamics in the social media trend related to a topic over time. Against this background, we propose a time-dependent analysis of the effect of textual characteristics on the volume of a discussion. As a major result, we find that the way users write their texts temporally anticipates the rise or the fall of a topic trend.