Artificial intelligence continues to revolutionize the landscape of online retail, transforming traditional shopping experiences into interactive digital engagements. Among the latest innovations are AI-powered digital streamers—virtual avatars designed to host livestream sessions and promote products to online audiences. Despite their futuristic appeal and potential cost advantages, a groundbreaking study conducted by researchers at the UBC Sauder School of Business reveals that these digital streamers currently underperform compared to their human counterparts, challenging widespread assumptions about the supremacy of AI in customer engagement.
The research rigorously analyzed sales data from a prominent fashion retailer operating on Tmall.com, one of the largest global e-commerce platforms. By comparing 328 distinct products before and after the introduction of digital streamers, the study established a clear performance hierarchy: human streamers led to substantial sales increases, digital streamers barely outpaced scenarios with no streamers at all, and the latter remained the baseline. This stark contrast illuminates the limitations of current AI streaming technologies and underscores the unique value human streamers bring—primarily, their authentic interaction and nuanced communication skills.
Delving deeper, the research team sought to uncover the underlying reasons for digital streamers’ underwhelming sales performance. Their subsequent experiments with an emerging online grocery retailer on the same platform introduced various iterations of digital avatars, ranging from simple cartoon-like characters to highly sophisticated human-like representations equipped with advanced features such as realistic voices and dynamic real-time interaction capabilities. This layered approach allowed the researchers to decode which aspects of digital streamer design most critically impact consumer response and purchasing behavior.
A pivotal finding emerged around the concepts of form realism and behavioral realism. The study distinguished form realism as the degree to which digital streamers visually resemble humans, while behavioral realism encompassed the streamers’ capacity to respond authentically and spontaneously to viewer inquiries during live sessions. Although improvements in visual fidelity and voice modulation contributed positively, the transformative breakthrough came with enhanced behavioral realism—specifically, the streamers’ ability to handle real-time question-and-answer exchanges with potential buyers.
Integrating real-time interactive Q&A functionality yielded an impressive 25 percent increase in units sold, alongside an 86 percent surge in revenue for featured products. This discovery highlights how dynamic and responsive engagement acts as a vital driver in bridging the gap between static virtual presentations and the electrifying immediacy characteristic of human-hosted livestreams. Furthermore, the introduction of a lottery mechanism, enabling audiences to win prizes during broadcasts, further amplified performance metrics by boosting sales by 17 percent and revenues by 70 percent, reinforcing the power of gamification strategies in digital commerce.
Beyond quantitative growth, these findings carry substantial implications for understanding consumer psychology in digital environments. Real-time interaction satisfies fundamental human desires for recognition and personalized attention, fostering trust and a sense of connection despite the artificial nature of the streamer. In contrast, less interactive or purely visual AI personas risk relegation to mere background noise, failing to captivate or convert viewers. The research thereby underscores a paradigm shift where AI’s effectiveness is not solely governed by its technical prowess but equally by its ability to simulate authentic social presence.
The study’s implications also extend to cost efficiency considerations for businesses. Unlike human streamers who require breaks, salaries, and have limited hours, digital streamers can theoretically function 24/7 without fatigue or wage demands. However, the current state of AI means cost savings alone won’t compensate for meaningful sales engagement unless interactivity is prioritized. These insights compel corporations to rethink their digital marketing investments and development roadmaps to focus on behavioral realism enhancements rather than superficial visual upgrades.
One particularly innovative proposition arising from the study suggests a hybrid model fusing AI and human strengths. By enabling humans to supervise multiple AI streamers simultaneously, companies could scale outreach while preserving the nuanced empathy and problem-solving acumen unique to humans. Such symbiotic collaboration could optimize workload distribution, elevate customer engagement, and provide instantaneous human intervention during complex queries, harnessing the best of both worlds in a cost-effective manner.
The UBC researchers caution that while AI-powered streamers are an emerging frontier, premature or unrefined deployment risks disappointing results that could hamper broader adoption. Their empirical evidence challenges the marketing myth that virtual avatars automatically guarantee enhanced sales metrics, instead positioning customer interaction as the essential differentiator. Consequently, efforts to refine neural language processing, speech synthesis, and computer vision must align with strategic user experience design to fully realize AI’s potential in live-stream commerce.
This pioneering investigation represents the first empirical foray into quantifying the influence of digital streamers on actual sales outcomes within real-world e-commerce ecosystems. By integrating rigorous data analysis with experimental design and simulation, the study offers actionable insights for both academia and industry stakeholders aiming to optimize the intersection of AI and retail. Moreover, it reframes the discourse from technological capability alone to encompass psychological engagement, user interaction, and strategic orchestration of virtual personas.
In conclusion, AI-powered digital streamers have undeniable promise as scalable, cost-effective solutions in the evolving online retail landscape. Yet, their success hinges fundamentally on the development of immersive real-time interaction capabilities that can rival the authenticity and responsiveness of human hosts. As artificial intelligence matures, fostering seamless human-machine collaboration and prioritizing behavioral realism will be critical to transforming digital avatars from mere novelties into powerful sales catalysts. Retailers and tech developers alike must heed these findings to avoid pitfalls and unlock the transformative potential of AI-driven livestream commerce.
Subject of Research: The empirical evaluation of AI-powered digital streamers’ effectiveness in online retail and design strategies to enhance their sales performance.
Article Title: AI-Powered Digital Streamers in Online Retail: Empirical Insights and Design Strategies from Experiments
News Publication Date: 26-Aug-2025
Web References:
Information Systems Research – Article DOI
Image Credits: Yanwen Yang
Keywords: Marketing, Business, Social sciences, Economics
