This article empirically confirms one core motivation for architectural zoning: Shape homogeneity among neighboring homes increases the value of residential buildings. Drawing on large-scale shape and transaction data, this study first develops a data-driven measure of architectural similarity, condensing three-dimensional shapes to univariate shape distributions. These algorithm-based similarity estimates are good predictors of human perceptions of shape similarity and are linked to property attributes and transaction prices. For the city of Rotterdam, a price premium of approximately 3.5% is estimated for row houses within very homogeneous ensembles over buildings facing heterogeneous neighbors.