The premise of massive multiple-input multiple-output (MIMO) is based around coherent transmission and detection. Majority of the vast literature on massive MIMO presents performance evaluations over simplified statistical propagation models. All such models are drop-based and do not ensure continuity of channel parameters. In this paper, we quantify the impact of spatially consistent (SC) models on beamforming for massive MIMO systems. We focus on the downlink of a 28GHz multiuser urban microcellular scenario. Using the recently standardized Third Generation Partnership Project 38.901 SC-I procedure, we evaluate the signal-to-interference-plus-noise ratio of a user equipment and the system ergodic sum spectral efficiency with zero-forcing, block diagonalization, and signal-to-leakage-plus-noise ratio beamforming. Our results disclose that at practical signal-to-noise ratio levels, SC channels yield a significant performance loss relative to the case without SC due to substantial spatial correlation across the channel parameters.