Pré-publication, Document de travail: This paper providesempirical support to the intuitive statement that urban geography matters to the success or failure of place-based public policies, using the French enterprise zone program as a case study. According to the few existing evaluations, this program has only had a small positive average impact on firm and job creation rates. In addition, this impact was shown to be strongly heterogeneous across the treated neighborhoods may account for part of these results. We estimate a series of augmented difference-in-differences models in which we interact the treatment indicator with a series of original indicators of spatial isolation, wich account for severance, peripherality and disconnection to transportation networks within the urban area. Results indicate that isolation does matter to explain spatial differentials in job creation and firm settlement rates across enterprise zones: only accessible neighborhoods were able to draw benefits from tax breaks and social exemptions. moreover, whereas the program mostly worked through a displacement effect on pre-existing firms, we show that urban geography was a clear determinant of the decision to create new firms from scratch.