Many policies are spatially targeted: they attempt to foster or incent policy actions in areas meeting specific criteria. The identification and delineation of the spatial reality of interest is a major challenge to spatially targeted policies. In many cases, this spatial reality has an ambiguous meaning and no coterminous governing institution, leading to the need for institutional collaboration around policies. This project investigates the discrepancy occurring when the functional space relevant for polices and the institutional space of government produced by historical processes do not share the same extent. The research examines two spatial realities of importance for public policies in the United States: metropolitan areas and disadvantaged communities. Because they are the main social unit of urban life, metropolitan areas—such as the San Francisco Bay Area—are the target of housing or economic development policies. Disadvantaged communities are targeted by programs intending to mitigate inequalities between communities. In both cases, specialists and laymen alike have a sense of what these spatial realities are, but charting a robust, actionable definition is problematic. The three articles of this doctoral project form a coherent path to scrutinize and resolve the spatial reality / institutional space disjunction. The first, *Identifying the space of urban life*, investigates spatial practices of individuals and how, collectively, they cast a unit of urban life, the metropolitan area. The second paper, *On point: designing robust spatial targeting for public policies*, analyzes how these practices can be integrated in a robust policy targeting, taking the example of disadvantaged communities in California. The third paper, *The place and scale of consent* takes a bottom-up approach to show how individuals perceive and express preferences for resolving that discrepancy. I show that neither of the currently used definitions for metropolitan areas and disadvantaged communities is robust because they use a strict threshold on a continuous metric that is not exhibiting clear breakpoints. Furthermore, definitions are not geographically consistent. A single definition does not capture the same reality of a metropolitan area in a coastal region or in the rural west, or of a disadvantaged community in a large urban region or in a rural county. Therefore, these definitions fail to provide a dependable policy target. Moreover, I show that local preferences for policy collaboration to resolve the discrepancy vary geographically, further undermining the relevance of a one-size-fits-all definition. I recommend that definitions of spatial targets for policies use a continuous scale instead of a discrete, binary one. These definitions should also be adapted to the geographical context they are applied in. Randomized vignette experiments can be used to understand residents' preferences.