The emergence of AI-generated virtual streamers in live-streaming commerce has complicated manufacturers' mode selection. In response, this study develops a game-theoretic decision-aid framework to analyze manufacturers' optimal strategies across both single live-streaming modes (self-streaming, streamer, and virtual streamer) and hybrid live-streaming modes (‘self−streaming+streamer’ and ‘self−streaming+virtual streamer’). It further incorporates different slotting fee structures and the platform's drainage effect to evaluate how these factors shape the manufacturer's mode choice. The findings show that: (i) consumer distrust may reduce demand for virtual streamers and the return on AI customisation, but a sufficiently strong drainage effect can offset this disadvantage; (ii) as streamer traffic effects strengthen, or drainage effects rise when traffic effects are still weak, the optimal strategy shifts from self-streaming to external or virtual streamer live-streaming, and eventually to hybrid modes; (iii) manufacturers prefer a single mode when both traffic and drainage effects are weak, but hybrid modes when these effects become stronger; and (iv) these results remain robust under different slotting fee structures and platform drainage strategies, while the manufacturer–platform equilibrium depends on drainage costs and off-peak incremental gains. Overall, the threshold-based framework guides manufacturers and platforms in allocating human and AI-enabled live-streaming resources across channels and periods.
Wu et al. (2026) studied this question.