Revenue estimation of telecommunication tower forms a key element in network planning, and many available methodologies tend to depend mainly on non-spatial factors with limited consideration for spatial heterogeneity. This study explores the integration of geospatial information in tower-level revenue modelling in an interpretable Multiple Linear Regression setting. Spatial factors that connected with demographic structures, socioeconomic status, the value of landed property, the attributes of the built environment and accessibility are also included to conventional non-spatial features. A wrapper-based forward stepwise procedure is applied for feature selection, and model validity is ensured through standard regression diagnostics and residual spatial dependence tests. The results show that incorporating spatial information leads to improved model behavior and more consistent prediction patterns compared to non-spatial baselines, with several spatial variables exhibiting statistically significant and stable regression coefficients. These results offer empirical support to the idea that using geospatial context, in addition to traditional attributes, increases the explanatory power and practical relevance of regression-based models for revenue driven network planning in heterogeneous regions. © 2026 IEEE.