Complex networks, and functional networks in particular, have become a standard tool to understand the structure and dynamics of real-world complex systems. One usually hidden assumption is that the structure of the reconstructed functional networks encodes useful information to guide interventions on the physical layer, when the latter is not known. We here test this assumption using a minimal model, simulating a propagation process in a physical network, and guiding interventions using node properties observed in the corresponding functional representation. We show how this approach becomes less optimal the more complex the topology is; up to becoming marginally better than choosing nodes at random in the real case of the European air transport network.