Praxeological turn in social research resulted in many examples of empirical studies using the concept of social practice. They are mostly case studies of a single or clearly defined practices, such as cooking, dancing or energy consumption. There is a lack of studies employing this concept as a framework for research on complex, dynamic and empirically hard to reach socio-spatial phenomena, such as urban housing estates. Drawing from the experience of a 4-year research project in six housing estates in three Polish cities we present and debate methodological challenges in applying theories of practice to urban research. We discuss methodological implications of representational and heuristic models of explanation -- which are being applied in practice-based research. We highlight the ambiguity of the notion of social practice in the existing literature and its troublesome 'double nature' -- implying that social practices are both implicit and observable phenomena --which, although alluring in the stage of conceptualisation, poses challenges in the stage of data collection, analysis and interpretation.