The Introduction of Human and AI-Powered Virtual Influencers on Social Media Marketing

In the rapidly expanding influencer marketing sector, positive consumer sentiment towards sponsored content plays a crucial role in the success of sponsoring brands and influencers. However, research shows that consumers are more likely to react negatively to sponsored videos than non-sponsored ones. We investigate how influencers can potentially diminish the negative impact of sponsorship on consumer sentiment by modulating their vocal characteristics in sponsored videos compared to non-sponsored videos. We use three causal identification strategies: instrumental variables, fuzzy regression discontinuity, and two-way fixed effects. To extract multi-modal 3V features (voice, visual, and verbal), we employ deep learning methods. We conclude that influencers significantly reduce the average loudness of their voices in sponsored videos relative to non-sponsored ones, and this difference in average loudness partly alleviates the negative impact of sponsorship on consumer sentiment. This vocal strategy of lowering loudness is more likely to be employed by influencers with shorter tenure. We discuss the implications for various stakeholders, including influencers, brands, consumers, and regulatory bodies.