Hedonic digital content backs a wide variety of business models. Yet, due to its experience good nature, consumers cannot assess its value before consumption. To overcome this obstacle, thumbnail images are frequently employed to provide an experience of content, and trigger views and sales. In spite of fragmented evidence from human-computer interaction research, thumbnails largely constitute a black box for research and practice. This research aims to fill this gap and asks: How and why do basic, conceptual and social features of thumbnail images affect popularity of hedonic digital content? To answer the question, we employ artificial intelligence imagery analysis to test and confirm a variance model against evidence from 400,000 YouTube videos. Our findings entail important theoretical contributions to visual perception in online contexts. In addition, this research proposes artificial intelligence imagery analysis as a new and fruitful research method for the largely visual information systems discipline.
Cremer, Stefan, "Predicting Popularity of Hedonic Digital Content via Artificial Intelligence Imagery Analysis of Thumbnails" (2017). PACIS 2017 Proceedings. 186.