Since the outbreak of the COVID-19 pandemic, the widespread use of online tools has diversified business communication strategies. However, there are few studies that look at the unique characteristics of these digital environments, especially when cameras are turned off. In this study, how the combination of visual images and voice expressiveness influences customer interest in online sales situations was investigated. Using quantitative and qualitative data, the change in customer interest for six video stimuli that combine different nonverbal cues from salespeople was analyzed: visual images (name only, photo without facial expression, smiling photo) and voice (with expression vs. without expression). The McNemar test revealed significant changes in customer interest for all combinations, with the combinations of expressive voice and smiling photo proving to be the most effective. The qualitative analysis with KH Coder’s Co-occurrence network showed that the characteristics of the voice (e.g., brightness and intonation) and the consistency between the visual image and voice are decisive for the customer’s impression. In particular, customers who initially showed interest tended to lose interest owing to discomfort and mistrust caused by monotone voices and the inconsistency of nonverbal cues, which prevented them from engaging with the content. Conversely, customers who initially showed no interest were persuaded by expressive voices and the consistency of nonverbal cues to become enthusiastic about the content and engage with it. These findings offer practical insights into digital communication strategies in sales, education, and advertising, particularly for customizing approaches based on initial audience interest.