Towards AI-based thumbnail design for fostering consumption on digital media platforms

Abstract

On digital platforms, thumbnails are omnipresent promotion tools for hedonic media goods. However, thumbnail designs still constitute a black box regarding their effects on consumption. To address this research gap, we apply artificial intelligence (AI) based imagery analysis to quantify and analyze almost 500,000 thumbnails drawn from two international digital media platforms – an advertisement-based video-sharing platform and a sales-oriented e-commerce one. Through multiple linear regression analyses, we provide quantitative evidence on the relationship between thumbnail designs and the consumption of hedonic media goods. For both digital media platforms, we find that several classes of visual cues and more faces especially with negative emotions foster consumption, while more text decreases consumption. We also find several distinct impacts on consumption across the two digital media platforms that we explain with the different underlying business models. We contribute to the literature on visual information processing and choice, AI-based design, and platform and e-commerce strategies. Further, we supplement the inventory of Information Systems (IS) research methods by adding AI-based imagery analysis. We provide implications for practice concerning AI-based thumbnail design and strategic pathways for digital platforms and offer directions for future research. We conclude with some key insights and a brief outlook.