An efficient and accurate approach for using diffusion Monte Carlo (DMC) to study molecular clusters is described and applied to an investigation of (H2O)(n) with n = 1, 2, 3. In this approach, importance sampled DMC is used to describe the intramolecular degrees of freedom of the water monomers. The intermolecular degrees of freedom are treated using standard unguided DMC approaches. This hybrid approach allows us to use significantly smaller simulations to obtain converged zero-point energies and ground state wave functions than would be required for a fully unguided simulation of the same system. We demonstrate the efficiency, accuracy, and utility of this approach through a study of nine isotopologues of water trimer, focusing on how the ground state probability amplitude for these clusters samples the various low-energy minima on the potential surface.