Show me your face: investigating the effect of facial features in review images on review helpfulness

Images generated by customers have become a critical component of product reviews. For fashion goods, some customers would embed review images in product reviews and disclose their faces when describing their product experiences, while how facial features affect other customers' perceived helpfulness of reviews remains largely unexamined. Drawing upon Information Adoption Model, this paper proposes that face disclosure and positive emotions revealed by facial expressions in product review images positively affect review helpfulness through increased credibility and emotion contagion effect. Specifically, deep convolutional neural networks are deployed to extract facial features from review images, and negative binomial model with product fixed effect is chosen to conduct empirical analyses based on a large-scale review dataset. We conducted propensity score matching to further deal with the selection problem, and the bias of coefficient caused by algorithm classification error is properly addressed. The empirical results and extensive robustness checks strongly support the positive effects of face disclosure and positive emotions. These findings enrich our understanding of how review images affect people’s information adoption behavior and provide viable guidance for visual content management on e-commerce platforms.