Tests of the concordance cold dark matter model on the scale of galaxies are so far inconclusive due to our poor understanding of the interplay between baryons and dark matter (DM). Through the construction of dynamical models that fit the observed motions of stars we can infer the total mass distribution. Stellar population models that fit the observed chemical properties of stars provide the stellar mass-to-light ratio needed to convert the observed light of stars into a stellar mass distribution. Subtracting the latter stellar (and if present gas) from the total mass distribution yields the distribution of dark matter. In this presentation, I show how combination of both dynamical and stellar population models allow us to disentangle the luminous and dark matter in galaxies of all types and of all masses from dwarf spheroidal to giant elliptical galaxies. At the same time, our chemo-dynamical modelling approach also provide insights into the formation history of galaxies, from a kinematically detected stellar stream as witness of a merger between dwarf galaxies, to the dynamical decomposition of giant galaxies into bulges and counter- rotating thin and thick disks.