Structural Equation Modeling of the Impact of AI-Powered Recommendation Systems on Consumer Behavior and Purchase Decisions in Nepalese Electronic Commerce

Abstract

This study examines the impact of AI-powered recommendation system attributes—personalization, accuracy, diversity, and transparency—on consumer purchase decisions in Nepal’s rapidly expanding e-commerce sector. The research aims to understand how these AI-driven features influence consumer browsing behavior, purchase intentions, satisfaction, and trust within a developing digital marketplace. A quantitative, cross-sectional research design was employed, involving 560 active online shoppers from diverse occupational and income groups across Nepal. Due to the absence of a comprehensive sampling frame, data were collected using convenience sampling through online platforms. A structured questionnaire was used to measure recommendation system attributes, consumer trust, and purchase decision outcomes. Confirmatory Factor Analysis (CFA) was applied to validate the measurement model, followed by Structural Equation Modeling (SEM) to test the hypothesized relationships. The findings reveal that AI-powered recommendation attributes have a significant and positive effect on consumer purchase decisions, with personalization emerging as the most influential factor, followed by accuracy and recommendation diversity. Consumer trust significantly moderates these relationships, strengthening their impact. However, excessive reliance on AI-generated suggestions may constrain product exploration, underscoring the need for transparent and diverse recommendation designs.