The concept of governance has provided many ways to theorise the shifting power relations between the state, interest groups and civil society over the last 30 years. Theorisations have culminated in ‘spatial imaginaries’ for visualising new governing practices and their associated power relations. By paying attention to these imaginaries, it is possible to see how each theory of governance brings particular spatialities of power to the fore, while necessarily foreclosing others. This foreclosure stems from a failure to visualise diverse and multiple modes of power in governance models and to take in power as a relative and spatially contingent property. This is not only theoretically significant, however; it also has important practical consequences for how we govern effectively in practice. I argue that rather than starting our analyses of governance arrangements with theoretical models which appear to predetermine our understanding of the spatial workings of power, we should instead remain open and attuned to the complex geographies of power that might actually operate in practices of governance on the ground. I suggest that by deploying John Allen's topological approach to power we can achieve a more relational and spatially contingent account of power in practice under the turn to governance. This will give us greater insight into actual governance arrangements and their limitations, exclusions and unevenness.