The ability to present effectively is essential for professionals; therefore, oral communication courses have become part of the curricula for higher education studies. However, speaking in public is still a challenge for many graduates. To tackle this problem, driven by the recent advances in computer vision techniques and prosody analysis, multimodal tools have been designed to support the development of public speaking skills. One of these tools is the Presentation Trainer, a research prototype able to provide learners with real‐time feedback on a set of nonverbal communication aspects. Despite initial positive evaluations, the application still lacks grounding in a valid assessment model for nonverbal communication aspects in the context of presentations. To come up with such a model, we conducted semi‐structured interviews with experts in the public speaking domain. Furthermore, the objective of these interviews was also to have a formative evaluation of the Presentation Trainer, analysing how it suits with common practices for teaching and learning public speaking skills. The results of this study identify 131 nonverbal communication practices that affect the quality of a presentation and summarize experts' points of view regarding multimodal public speaker instructors.