AI-Enabled Demand Forecasting Capability, Inventory Resilience, and Operational Performance among Vietnamese Retail SMEs: The Mediating Role of Decision Quality

Abstract

Retail SMEs across Southeast Asia increasingly experiment with AI-enabled analytics to improve demand forecasting, yet performance outcomes remain uneven because technology adoption does not automatically translate into better operational decisions. This study examines how AI-enabled demand forecasting capability influences operational performance among Vietnamese retail SMEs through inventory resilience and decision quality. Building on capability theory, information processing logic, and resilience perspectives, the model conceptualizes AI capability as a set of routines that combine data readiness, tool reliability, and managerial interpretive competence. Survey data were collected from 455 Vietnamese retail SMEs that had used digital analytics tools for forecasting or replenishment planning for at least six months. Partial Least Squares Structural Equation Modeling (PLS-SEM) was used to test direct effects and parallel mediation. Results indicate that AI-enabled forecasting capability is positively associated with operational performance, with decision quality and inventory resilience operating as complementary mediators. Decision quality strengthens resilience by improving replenishment timing and reducing overreaction to short-term noise. The findings clarify why AI initiatives often fail when data and interpretation routines are weak, and they offer ASEAN-relevant insights for small retailers facing demand volatility and supply disruptions. Practical implications emphasize pairing analytics tools with data governance and managerial training, enabling SMEs to convert forecasts into robust inventory actions.