The Power of Beauty: Leveraging on Deep Learning to Disentangle the Effect of Image Aesthetics on Online Purchase

Visuals have been deemed as one of the primary marketing formats in service e-tailing environ-ments to compensate for the abstract and intangible nature of services. Yet, despite the overwhelm-ing empirical evidence attesting to the importance of aesthetics in product and website designs, there is a dearth of research on how the beauty of images depicting service offerings could affect consumers’ behaviour. Subscribing to Stimuli-Organism-Response (S-O-R) framework, we attempt to elucidate the power of aesthetics in shaping consumers’ cognitive and emotional responses when they are exposed to the portal image of a service offering. Additionally, we endeavour to explore the heterogeneity of consumers’ responses or their centrality of visual aesthetics, to such images. Blending computer vision and deep learning techniques, we advance a computable and decomposable aesthetic assessment method for scoring the aesthetics of portal images belonging to more than 299,000 local service offerings on a leading Chinese group buying site. We then em-ploy Propensity Score Matching (PSM) to yield initial evidence demonstrating that aesthetics ex-erts significant effects on the online sales of services.