The description of the electronic structure of the transition metal compounds is complicated by the localization of the d orbitals. The first breakthrough came with Hubbard's proposal of a charge fluctuation model that included localized, correlated d orbitals and the possibility of electron hopping from one metal site to another. However, it was realized that in order to have an accurate understanding of some compounds it was necessary to include the possibility of charge transfer from the ligand to the metal site. This gave rise to the Zaanen-Sawatzky-Allen (ZSA) model for transition metal compounds. Experimental proof of this model is hampered because the traditional tools, photoemission and inverse photoemission, require experimental conditions (single crystals, ultrahigh vacuum) that are not always met. In this talk an alternative means to test the ZSA model using soft x-ray absorption and emission at both the transition metal L edge and the ligand K edge will be discussed. It is shown that the ligand K emission allows an unambiguous determination of the origin of the top of the metal band, and therefore of the nature of the electronic gap. This procedure is used to investigate the Mott-Hubbard to charge transfer transition of the transition metal difluorides (MF2, M= Cr - Zn). We found that, except for the filled Zn ion, all the top valence bands are predominantly metal 3d orbitals hybridized with fluorine 2p orbitals. Therefore, all compounds in this transition metal difluoride series are Mott-Hubbard insulators. By extrapolating our results we accurately reproduce the well known behavior of the transition metal monoxides. More examples dealing with transition metal ternary compounds will also be presented.