Small and Medium Industries (SMIs) play a crucial role in regional economic development, particularly in metropolitan areas such as Surabaya, Indonesia. Nevertheless, the high heterogeneity of SMI characteristics poses challenges for designing effective and targeted development policies. This study proposes a data-driven clustering framework to identify patterns and characteristics of SMIs in Surabaya by employing the Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) algorithm. The novelty of this research lies in the application of BIRCH for large-scale SMI data clustering combined with systematic parameter evaluation using the Silhouette Score to ensure cluster quality and stability. BIRCH was chosen for its efficiency in handling large and heterogeneous datasets through hierarchical summarization, addressing limitations of methods such as K-Means that require predefined cluster numbers and are sensitive to initial centroids. The dataset was preprocessed through missing value handling, data type transformation, categorical label encoding, and numerical standardization. After preprocessing, 31,472 records with six variables were analyzed. Various combinations of threshold and branching factor parameters were evaluated using the Silhouette Score to determine the optimal configuration. The best result was obtained with a threshold of 0.7 and a branching factor of 50, achieving a Silhouette Score of 0.743 and forming five distinct clusters. The resulting clusters exhibit clear structural patterns in terms of land area, initial capital, labor force, business scale, company type, and risk level. The findings demonstrate that BIRCH effectively produces well-separated and interpretable clusters, providing a robust analytical basis for evidence-based policymaking in SMI development.