AI-driven consumer insights in live E-commerce : a structural equation modelling approach

Live e-commerce has fundamentally transformed online retail, yet analysing the massive volume of real-time data it generates remains a significant challenge. While existing research has explored various facets of live streaming, the specific impact of AI-driven consumer insights on purchase intentions remains underexplored. This study investigates this relationship within the context of sports apparel live streams on the Douyin platform. We propose and validate an integrated framework that combines Large Language Model (GPT-3.5) analysis with Structural Equation Modelling (SEM) to examine the pathways linking AI-driven insights, perceived informational value, platform trust, and purchase intentions. Analysis of 200 live streaming sessions reveals that AI-driven insights significantly enhance consumer engagement (by 42.5%) and conversion rates (by 24.3%). The framework demonstrates strong construct validity and excellent model fit (CFI = 0.942). These findings provide retailers and platform operators with a validated roadmap for implementing AI-driven strategies to optimize customer engagement and sales performance.