The Influence Of AI-Driven Marketing Communication Through Customer Trust On Purchase Decision Of Product In Indonesia Marketplace
Description
The integration of Artificial Intelligence (AI) in digital marketing has revolutionized the e-commerce landscape, offering highly personalized consumer experiences while simultaneously provoking the personalization-privacy paradox. This study examines the structural relationships between AI-driven marketing communication, Electronic Word of Mouth (eWOM), customer trust, and consumer purchase decisions within the Indonesian marketplace ecosystem. Employing an explanatory research design, a quantitative approach was executed via an online survey targeting 400 active e-commerce users selected through purposive sampling. Data analysis was conducted using Partial Least Squares Structural Equation Modeling (PLS-SEM) via SmartPLS. The empirical findings reveal that all seven hypotheses are positively and significantly supported. AI-driven marketing communication and eWOM exert substantial direct impacts on both customer trust and purchase decisions. Crucially, customer trust emerges as the most dominant direct predictor of final purchase decisions and acts as a significant partial mediator that bridges both digital interactions (AI-marketing and eWOM) toward transactional conversion. These results underscore that technological sophistication and social proof require a foundational framework of trust to successfully mitigate perceived risks and uncertainties. Conceptually, this research enriches contemporary consumer behavior literature by integrating algorithmic automation and social confirmation into commitment-trust theory. Practically, e-commerce operators are urged to balance algorithmic efficiency with heightened transparency, data privacy ethics, and review authenticity to cultivate robust, sustainable consumer trust.
Files
Steps to reproduce
Employing an explanatory research design, a quantitative approach was executed via an online survey targeting 400 active e-commerce users selected through purposive sampling. Data analysis was conducted using Partial Least Squares Structural Equation Modeling (PLS-SEM) via SmartPLS. The empirical findings reveal that all seven hypotheses are positively and significantly supported