Abstract

On streaming platforms with vast and diverse content, digital advertising—especially via social media—has become increasingly important alongside traditional promotional methods in movie marketing. Grounded in Dual Coding Theory, we hypothesize that congruence between visual elements (e.g., posters) and verbal elements (e.g., synopses) in digital ads positively affects advertising effectiveness. Prior research has largely relied on subjective human judgments to assess image–text congruence, limiting measurement objectivity. This study proposes a data-driven, objective approach to measuring image–text congruence and empirically examines its impact on customer engagement. Using IMDb data, we employed LLaVA, a multimodal model, to generate textual descriptions of movie poster images and computed cosine similarity between these descriptions and the corresponding synopses, thereby operationalizing an image–text congruence score. Extending the Theory of Cognitive Consistency, we also test the effects of consumer–ad imagery congruence and brand–ad congruence on brand evaluations, incorporating image–text congruence as a key factor. The importance of image–text congruence is further validated via variable-importance analyses from machine-learning models and a consumer survey. Based on these findings, we offer theoretical and practical implications for optimizing movie marketing strategies and enhancing customer engagement.