Purpose: This study investigates how perceived usefulness influences TikTok Shop users’ continuance intention, with attitude and satisfaction serving as mediating variables. It aims to clarify the psychological and technological mechanisms driving user retention, particularly among Indonesian Generation Z and MSMEs. Methodology/approach: A quantitative survey design was applied, involving 420 purposively selected TikTok Shop users in Yogyakarta and surrounding areas. Data were analyzed using Partial Least Squares-Structural Equation Modeling (PLS-SEM) to evaluate both measurement and structural models. Validity and reliability tests, along with mediation analysis, were conducted to assess hypothesized relationships. Results: Findings show that perceived usefulness significantly enhances user attitude, satisfaction, and continuance intention. Attitude has a strong positive effect on continuance intention, whereas satisfaction, despite being positively related, is not a significant direct predictor. Mediation analysis indicates partial mediation through attitude and weaker mediation through satisfaction, underscoring the stronger role of cognitive evaluations over emotional responses in driving platform loyalty. Conclusions: The study concludes that user retention on TikTok Shop is primarily shaped by perceived usefulness and positive attitudes rather than satisfaction alone. Enhancing platform utility and fostering favorable user perceptions are critical for sustaining engagement. Limitations: The cross-sectional design, geographic focus on Yogyakarta, and reliance on purposive sampling limit generalizability. Contextual factors such as culture or platform features may also moderate the findings. Contribution: The research integrates UTAUT2 and Expectation-Confirmation Theory, advancing theoretical understanding of social commerce continuance intention. Practically, it offers actionable insights for developers and marketers to optimize features, enhance perceived usefulness, and build favorable attitudes to retain users.