U-shaped Effect of Color in Short-form Video: Combination of Text, Audio, Image ML and Generative AI

Short-form video has revolutionized digital marketing. While extensive research has focused on image color, the impact of video color on short-form video marketing remains underexplored. In this study, we used machine learning techniques to analyze 2,984 videos and develop a multimethod framework organized into two stages across 4 studies. In the first stage, we explore how video color at cover level impacts customer’s click (Study 1). In the second stage, following the customer click, we investigate how video color at peak level impacts customer’s purchase behavior (Studies 2–4). Notably, Study 2 explores the U-shaped relationship between video color at peak level and sales. Furthermore, Study 3 investigates how audio arousal and text sentiment moderate this effect. To further provide causal evidence, Study 4 uses generative AI to design video stimuli and conducts 2 randomized online experiments, replicating the empirical results of before. Theoretically, these findings expand the literature on color and short-form videos marketing. Methodologically, this research provides new directions for leveraging machine learning techniques in video analysis for marketing and exploring the application of generative AI in marketing strategies.