Implementation of AI-Based Chatbot in Marketing Service Information System in Retail Companies

Abstract

The proliferation of AI-driven retail platforms in Indonesia has outpaced empirical understanding of how cultural factors shape their adoption and effectiveness. This study addresses a specific gap: existing hybrid TAM–SERVQUAL models treat cultural variables as contextual background rather than structural moderators, limiting their explanatory power in high-context, collectivist markets. Specifically, this research examines whether interpersonal trust, collectivism orientation, and relation-based communication preferences significantly moderate the relationships between AI-based marketing information systems and customer satisfaction, and whether these moderation effects differ between large retailers and micro, small, and medium enterprises (UMKM) in Indonesia. To fill this gap, we develop and formally validate the Culturally-Moderated AI Adoption Framework (CMAAF), which extends TAM and SERVQUAL beyond additive integration by formally specifying cultural boundary conditions that amplify or attenuate AI’s effect on satisfaction. Unlike prior hybrid models, CMAAF is validated here using confirmatory factor analysis (CFA) and structural equation modeling via SmartPLS 4.0 with a sample of 347 respondents across five major Indonesian cities. Reliability ranged from α = 0.782 to 0.891. Results confirm that AI personalization (β = 0.421), recommendation systems (β = 0.318), and chatbots (β = 0.267) significantly predict customer satisfaction (R² = 0.637), with cultural variables acting as significant structural moderators, particularly in the UMKM segment. These findings demonstrate that CMAAF provides a more precise and culture-sensitive predictive framework than existing hybrid models, with direct implications for AI deployment strategies in emerging economies.