Analyzing the impact of faces on consumer engagement in social media videos: a machine learning approach

Purpose This study aims to gain insight into the use of the human face in different contexts in social media videos and its impact on consumer engagement with brands. Design/methodology/approach Using machine learning techniques, this research collected 2,754 social media videos posted by the top 50 brands across a three-year time frame. Data were analyzed using generalized estimation equations. Findings Consistent with attachment theory, the findings suggest that the mere presence of the human face in social media videos has a robust, positive effect on consumer engagement, which further increases with larger face sizes. However, the results also show boundary conditions with face duration on-screen and hedonic (vs utilitarian) brand positioning. Research limitations/implications This study focused on top brands posting videos on TikTok. This study should be replicated in other contexts such as additional platforms and brands. Practical implications Managers trying to grow their brand using social media videos should feature human faces (and closer-up human faces), with varying effects if the goal is comments versus shares versus likes. Originality/value This research examines how faces impact user engagement with social media videos. It provides insights into how a brand-related variable (hedonic vs utilitarian) can moderate the impact of faces and further expands the learnings of attachment theory in branded social media videos. Methodologically, a key part of this is the scalability of machine learning to code videos with human faces at a frame-by-frame level.