<p>This paper introduces an innovative approach for constructing large-scale, realistic benchmark geospatial datasets specifically designed for the single-allocation hierarchical hub median problem, a key variant of the hierarchical hub location problem. Addressing the lack of publicly available large-scale datasets in location science, we have provided a detailed step-by-step methodology to generate benchmark datasets. Also, we have provided datasets of various granularities based on Kolkata and Mumbai, two of India’s most populous metropolitan cities as a ready reference. The datasets are built using actual building-level geographic data acquired from OpenStreetMap and QGIS. Benchmark datasets are solved for the single-allocation hierarchical hub median problem formulation using IBM ILOG CPLEX to provide exact solutions, supporting robust empirical validation. The datasets are fully reproducible, scalable, and adaptable to various hierarchical hub location problem formulations and urban planning scenarios. By making these resources publicly available, this work fills a significant research gap and offers a foundational platform for algorithmic benchmarking and future methodological innovations in hierarchical hub location studies.</p>