As an emerging marketing tool, virtual influencers (VIs) have attracted increasing scholarly attention. However, existing research predominantly adopts linear causal analysis models, which fail to capture the complex, nonlinear interaction between consumers and VIs. Grounded in the 5W communication model and utilizing a fuzzy-set qualitative comparative analysis (fsQCA), this study systematically explores how different configurational paths influence consumer engagement, drawing on empirical data from 205 participants. The findings reveal that (1) the synergy of entertainment, information, and credibility is a core prerequisite for achieving high engagement; (2) two equivalent paths—namely, the technology-driven path (media richness + content synergy) and the cognition-driven path (technology acceptance + content synergy)—lead to high engagement, both with a solution consistency of 0.98; and (3) the joint absence of content, media richness, and audience cognition results in low engagement. Theoretically, this study challenges traditional linear approaches by validating causal asymmetry and revealing configurational interdependencies among communication elements. It also extends the Media Richness Theory (MRT) and the Technology Acceptance Model (TAM) into the context of virtual influencer (VI) marketing. Practically, the proposed dynamic configuration model offers marketers a novel framework for optimizing VI campaigns through resource-adaptive strategies.