X-ray powder diffraction (XRPD) analysis of intact multi-component consolidated mixtures has significant potential owing to the ability to non-destructively quantify and discriminate between solid phases in composite bodies with minimal sample preparation. There are, however, limitations to the quantitative power using traditional univariate methods on diffraction data containing features from all components in the system. The ability to separate multi-component diffraction data into patterns representing single constituents allows both composition as well as physical phenomena associated with the individual components of complex systems to be probed. Intact, four-component compacts, consisting of two crystalline and two amorphous constituents were analyzed using XRPD configured in both traditional Bragg–Brentano reflectance geometry and parallel-beam transmission geometry. Two empirical, model-based methods consisting of a multiple step net analyte signal (NAS) orthogonalization are presented as ways to separate multi-component XRPD patterns into single constituent patterns. Multivariate figures of merit (FOM) were calculated for each of the isolated constituents to compare method-specific parameters such as sensitivity, selectivity, and signal-to-noise, enabling quantitative comparisons between the two modes of XRPD analysis.