Investigating consumers’ cognitive, emotional, and behavioral engagement in social media brand pages: A natural language processing approach

Abstract

By applying a natural language processing approach, the present study aims to reveal what features of brand posts (i.e., post type, vividness, interactivity, and valence) may impact consumers’ behavioral, cognitive, and emotional engagement respectively. After collecting brand posts data from a microblog website, multilevel mixed-effects regression models were utilized to analyze the panel data set. It was found that interactive strategies, such as using interactional posts and integrating interactivity cues, are effective in stimulating all three dimensions of consumer engagement. Post valence, or post sentiment, is positively associated with consumers’ behavioral and emotional engagement. Nevertheless, using photos or videos is found to be less effective in motivating consumer engagement. Among these three dimensions of consumer engagement, cognitive engagement is more difficult to be stimulated compared with behavioral and emotional engagement. Implications of the findings are discussed at the end of the paper.