As the closure rate of small and medium-sized enterprises (SMEs) in Korea rises, there is a growing need for models that can predict closure risk in advance. Prior studies rely heavily on financial statements that are often unavailable for SMEs, limiting their practical relevance. This study addresses this gap by using online consumer behavior data as a proxy for SMEs' business value and integrating them into a Distance-to-Default (DTD)–based structural model to assess closure risk, while accounting for regional heterogeneity. Using SME-level data, the proposed approach significantly outperforms existing methods, increasing the area under the curve (AUC) by 0.20 and recall by 21.8%. Long-term risk level, short-term risk trend, and regional cost conditions emerge as the most important predictors, and the model achieves a recall of 80.3% for four-week-ahead predictions. Overall, this study presents an online consumer data based early warning system that offers practical implications for policymakers and financial institutions.