Understanding urban crime dynamics is an enduring challenge in urban planning and criminology. The environmental approach to security – with environmental criminological theories and CPTED – gives a direction emphasizing the interplay between built form and opportunity structures for crime. Within this framework Space Syntax prove helpful in modelling people behaviour and exploring social dynamics. In this perspective, the recent Spatial Crime Impedance (SCI) hypothesizes that configurational indices combinations could be used to understand natural predisposition of places in attracting or repelling certain types of crime. However, this theoretical construct remains underexplored and requires empirical testing. In this vein, the present study proposes a novel methodological framework to operationalize and assess SCI by combining spatial configuration analysis and supervised machine learning to assess whether different crime types—street robbery, residential burglary, and drug dealing—can be associated with distinctive configurational properties in the case of Pisa, Italy. Initial results show a modest predictive potential and suggest that while spatial configuration alone may not fully explain crime distribution, it contributes non-negligible predictive power. This study offers a first step toward operationalizing SCI, highlighting the relevance of spatial form in evidence-based strategies for urban crime prevention and spatial justice.