Direct-detection experiments for light dark matter are making enormous leaps in reaching previously unexplored model space. Several recent proposals rely on collective excitations, where the experimental sensitivity is highly dependent on detailed properties of the target material, well beyond just nucleus mass numbers as in conventional searches. It is thus important to optimize the target choice when considering which experiment to build. We carry out a comparative study of target materials across several detection channels, focusing on electron transitions and single (acoustic or optical) phonon excitations in crystals, as well as the traditional nuclear recoils. We compare materials currently in use in nuclear recoil experiments (Si, Ge, NaI, CsI, CaWO4), a few of which have been proposed for light dark matter experiments (GaAs, Al2O3, diamond), as well as 16 other promising polar crystals across all detection channels. We find that target- and dark-matter-model-dependent reach is largely determined by a small number of material parameters: speed of sound, electronic band gap, mass number, Born effective charge, high-frequency dielectric constant, and optical phonon energies. We showcase, for each of the two benchmark models, an exemplary material that has a better reach than in any currently proposed experiment.