The spatial mismatch hypothesis argues that low-skilled black residents, who are restricted to excluded ghettos, have been isolated from the knowledge of job opportunities by the suburbanisation of jobs. The result of this emerging spatial mismatch is higher rates of unemployment among low-skilled black workers. Research on this question usually relies on a ‘deductive’ methodology in which causal mechanisms are proposed and then tested using sample surveys. The logic of this argument follows the deductive-nomological model of explanation in which statistical associations are established between independent and dependent variables. We argue that this type of explanation has the characteristics of a ‘black box’ explanation because it ignores the social mechanisms that may cause certain outcomes. As an alternative, we propose a ‘realist’ approach to the study of the labour market spatial mismatch that uses a ‘retroductive’ methodology to discover causal mechanisms. By using this approach, we have established that the residents of excluded ghettos in Cape Town are not necessarily isolated from information on the suburban job market. Through a variety of workplace mechanisms, workers create a wide range of social networks that extend well beyond the confines of their neighbours and reach into networks of both employers and colleagues.