Purpose This study aims to examine how the Chinese climate nongovernmental organization “Chinese Weather Enthusiasts” engaged youth through video strategies. Design/methodology/approach The research proposed a framework grounded in the 5W model and message sensation value (MSV) to analyze the relationship between video content and user interaction. It categorized Bilibili videos into outer and inner features and introduced rhetorical strategies as content elements. A hybrid video coding framework was used, combining machine learning and deep learning (computer vision) for analyzing formal features, while manual coding was used for content features. Findings The results revealed that video length, long shots and the number of scenes positively influenced coins and favorites, whereas personification had a negative impact. In addition, tone and language intensity were positively correlated with user engagement. Originality/value This study offers insights regarding video production for climate communication, broadening the focus from text and images to video content and providing evidence-based guidance for practitioners.