Using sentiment analysis, k-means clustering, and latent Dirichlet allocation (LDA), we collected and analyzed more than 50,000 user-generated content (UGC) related to five athletic apparel brands, to examine the content characteristics in brand-related UGC from three types of social media: Twitter (self-media), Reddit (collaboration platform), and Instagram and YouTube (creative outlets). We find that UGC on creative outlets contains more positive brand sentiments and emotional trigger words. UGC on self-media and collaboration platforms are brand-centered, while the most popular topics on self-media are brand news/trends, and it is a product-related question on collaboration platforms. UGC on creative outlets are self-promoting, while creators’ motivations and achievements are the most popular on Instagram, and activity-related content and informative reviews are the most popular on YouTube. Besides, the formation of brand communities positively influences consumer engagement on social media. This research extended from the previous framework for comparing brand-related UGC with big data and machine learning techniques. It provides insights into how channel characteristics result in differences in brand-related UGC, which are valuable to future researchers and brand managers.