Abstract Dust aerosols comprise irregular grains and aggregates with various mineralogical mixtures. However, most optical models use effective refractive indices to represent particles. This study develops an inhomogeneous aggregate model assigning mineral‐dependent refractive indices to constituent elements and representing global mineral volume fractions. A synergistically unified approach combining the invariant‐imbedding T‐matrix, improved geometric optics, and physical‐geometric optics methods computes the single‐scattering properties. Furthermore, polarized Monte Carlo radiative transfer simulations are conducted to quantify lidar observables. Mineralogical inhomogeneity has little effect on the extinction efficiency but substantially affects the single‐scattering albedo, asymmetry factor, backscatter, lidar ratio, depolarization, and phase‐matrix elements, with pronounced impact for effective radii above approximately 1 μm. Simulated depolarization and attenuated backscatter are higher for inhomogeneous aggregates than the homogeneous counterparts. Homogeneous effective‐medium models can preserve extinction while introducing biases in polarization and lidar‐sensitive properties, with downstream implications for dust optical and microphysical retrievals.