Predicting brand confusion in imagery markets based on deep learning of visual advertisement content

In the consumer goods industry, unique brand positionings are assumed to be the road to success. They document product distinctiveness and so justify high prices. However, as products are getting more and more interchangeable, brand positionings must rely-at least partially-on supporting advertisements. Here, especially ads with visual content (e.g. photos, video clips) are able to connect brands with desirable emotions and values. Recently, besides TV, cinema, newspaper, also search engines, social networks, photo-, video-sharing platforms are used to spread such ads. In this paper, we demonstrate, how deep learning based on such ads can be used to predict uniqueness of brand positionings. A sample application to the German Pils beer market is used for demonstration.