In this work, we propose a new methodology for mixture design. By projecting the problem on the space of individual component properties, the methodology exploits a natural problem decomposition and capitalizes on fast methods for pure compound design and mixture fraction design. We demonstrate the proposed methodology through application to two illustrative examples and then to two problems from the mixture design literature concerning the purification of ibuprofen. In all cases, the proposed approach finds optimal solutions while exploring a small number of feasible molecular structures. © 2016 American Institute of Chemical Engineers AIChE J, 62: 1514–1530, 2016