Locating the peaks of solutions via the maximum principle: I. The Neumann problem