Artificial Intelligence–Driven Cross-Platform Data Integration and Its Impact on Consumer Purchase Intention: The Mediating Role of Personalization effectiveness and Consumer Trust in AI

Abstract

Despite its prominent role in digital commerce, the impact of AI on actual purchasing behavior is under-theorized and understudied. There is a theoretical and empirical gap in understanding the impact of artificial intelligence on the actual purchase by consumers. This study constructs and tests an integrated model based on Stimulus-Organism-Response (S-O-R) and Trust Theory to illustrate the relationships among four AI capabilities (data integration, algorithmic recommendation quality, transparency, and perceived intelligence), purchase intention, mediated by personalization effectiveness and trust in AI, in the context of cross-platform e-commerce. 229 Bangladeshi consumers were digitally active and these were analyzed with the help of PLS-SEM (SmartPLS 3.2.8). Measurement model had good reliability (composite reliability 0.867-0.938; Cronbach's α 0.796-0.912) and validity (AVE 0.620-0.791; HTMT<0.85). In contrast to the hypothesized models, the four types of AI did not directly predict purchase intention. On the other hand, three indirect paths were found: recommendation quality influenced purchase intention via personalization effectiveness (β = 0.144, p = 0.031); while transparency influenced purchase intention via trust in AI (β = 0.151, p < 0.001) and perceived intelligence influenced purchase intention via trust in AI (β = 0.186, p < 0.001). The model explained 55.6% of the variance in purchase intention, 46.3% in personalisation effectiveness and 32.9% in trust in AI. Combined, these findings indicate that instead of being valuable in itself, the value of AI skills is more likely to stem from the consumer's reception and mental evaluation of the technology. The study offers insight for both AI-consumer behavior research and for e-commerce firms aiming to get better returns on AI investment by prioritizing personalization and trust.