Online reviews are a step to communicate various brands or products. Product reviews usually contain product features, advantages, and disadvantages, as well as other products for comparison. Electronic product reviews conducted online on YouTube channels are the object of this research. With certain criteria, five videos were obtained as review objects that will be studied in this research. The 458 dataset of audience comments on this review will be studied qualitatively using machine learning in the form of software of Nvivo, as well as internet ethnography and pragmatics. From the analysis results, it is known that each electronic product review video provides positive, negative or neutral perceptions and attitudes from the audience. Audience sentiment is formed from product features, images of reviewers, how it was reviewed, and product image. By the limitation of the research, it needs to be developed further with different methods and objects to provide more input to producers or reviewers.