Strategies to Enhance Perceived Usefulness and Ease of Use: Impact on E-Commerce Platform Usage Intention

This study develops an integrated cognition–psychology–behavior framework to examine how subjective norms (SN) are internalized into cognitive beliefs through the mechanism of consumer trust (CT), thereby influencing behavioral intention. Using data collected from a stratified random sample of 896 e-commerce users in Taiwan, the findings address a key limitation of the traditional Technology Acceptance Model (TAM) by explicating the process through which social influence is transformed into psychological evaluation. Covariance-based structural equation modeling (CB-SEM) was employed for hypothesis testing, and PROCESS macro analysis was conducted to assess conditional indirect effects, enhancing the robustness of the mediation and moderation results.The results indicate that perceived usefulness (PU) and perceived ease of use (PEOU) both have significant positive effects on usage intention (UI). Consumer trust emerges as the strongest predictor of intention (β = 0.270, p < 0.001) and serves as a critical mediator in both the PU/PEOU → UI and SN → UI relationships. In addition, subjective norms exert both direct (β = 0.255, p < 0.001) and indirect effects on usage intention through their influence on consumer trust.Regarding AI technical characteristics, the findings show no significant direct effect on usage intention (β = 0.023, n.s.), suggesting that these features do not function as primary determinants. Instead, they act as contextual moderators: AI transparency, personalization, and interactivity strengthen the effects of cognitive evaluations (PU and PEOU) on usage intention, thereby serving as important boundary conditions.Overall, the findings suggest that in intelligent e-commerce contexts, improvements in system functionality and usability should be complemented by trust-building mechanisms and social influence strategies to enhance behavioral outcomes. Furthermore, AI design should emphasize transparency and controllability to facilitate the translation of cognitive evaluations into actual usage behavior. These findings provide practical implications for platform design, algorithm governance, and future longitudinal and field-based research.