Visual-centric social media platform has rapidly grown as a marketing tool and image search engine and has received attention from many users and various brand marketers. However, clickbait, which mainly originated in the form of news headlines, has appeared in visual social media in the form of discrepancies between images and the accompanying text. This clickbait can lead to biased or inefficient information retrieval for users and even a variety of problems for businesses attempting to increase their brand awareness and sales opportunities through their activities in visual social media. Although existing methods identify clickbait via reports from users and automated algorithms based on text media, little research has focused on visual clickbait in visual social media platforms, such as Instagram. In this work, we examine user’s perception on clickbait images to identify visual informativeness to develop a novel approach for characterize and detect clickbait in visual social media with a focus on the topic of fashion. By integrating different types of features, we were able to classify clickbait with an accuracy of 0.864 and categorized five types of clickbait images: scenery, graphics, products, cosmetics and food. Additionally, proposed brand-specific models were applied to three high-end couture brands, namely, Cartier, Chanel, and Hermes, which are often mentioned in clickbait posts. This approach shows that the classification of specific brand-related posts by the brand-specific models was 20% more accurate on average than the clickbait classification of the general model. Furthermore, we confirmed the potential use of the proposed model by examining the impact of the clickbait classification on the brand’s perception.