Exploring the role of virtual influencers and digital twin technology in tourist decision-making: a PLS-SEM and multilayer perceptron analysis

Purpose To examine how virtual influencers (VIs) and digital twin (DT) technology shape tourist decision-making, engagement, and destination loyalty, and to test the moderating role of perceived authenticity (PA) in tourism marketing, drawing on social influence theory (SIT) and the elaboration likelihood model (ELM). Addressing this key gap, we articulate a complementary-process view in which VI social cues (SIT) prime attention, while DT diagnosticity (ELM) provides central arguments, jointly influencing persuasion. Design/methodology/approach Survey data from 619 respondents were analyzed using partial least squares structural equation Modelling (PLS-SEM) to estimate direct, indirect, and moderating effects, complemented by a Multilayer Perceptron (MLP) neural network to capture non-linear relationships. We performed sequence analyses as PLS-SEM→MLP to (1) validate the theory-driven model and (2) assess predictive, potentially nonlinear patterns. Model performance was assessed with explained variance (R2), without MLP performance for tourist decision-making reaching up to 0.98. Findings Digital twin technology significantly enhances tourist engagement and destination loyalty. The perceived authenticity of virtual influencers is a key driver of tourist decision-making. Distinguishing source-level VI authenticity from message/content authenticity reduces construct redundancy and clarifies the mechanisms. PA moderates the effectiveness of both VI and DT initiatives, altering their impact on engagement, loyalty, and decision outcomes. The MLP results reveal strong nonlinear effects among the constructs, corroborating and extending the PLS-SEM findings. Practical implications Tourism marketers can leverage influences with high perceived authenticity and deploy digital twin experiences to boost traveler engagement and destination loyalty. Integrating both technologies strategically enhances the persuasive impact and improves decision-making across digital touchpoints. Originality/value We integrate SIT and ELM through a VI-as peripheral frame/DT-as central evidence complementarity, revealing engagement as a mediator between cues and loyalty. Our innovative PLS-SEM with nested-CV MLP + SHAP methodology uncovers nonlinear relationships, offering a mechanism-focused framework that advances AI use in tourism analytics.