Mid-gap luminescence in copper (Cu+)-doped semiconductor nanocrystals (NCs) involves recombination of delocalized conduction-band electrons with copper-localized holes. Silver (Ag+)-doped semiconductor NCs show similar mid gap luminescence at slightly (similar to 0.3 eV) higher energy, suggesting a similar luminescence mechanism, but this suggestion appears inconsistent with the large difference between Ag+ and Cu+ ionization energies (similar to 1.5 eV), which should make hole trapping by Ag+ highly unfavorable. Here, Ag+-doped CdSe NCs (Ag+ :CdSe) are studied using time-resolved variable-temperature photoluminescence (PL) spectroscopy, magnetic circularly polarized luminescence (MCPL) spectroscopy, and time dependent density functional theory (TD-DFT) to address this apparent paradox. In addition to confirming that Ag+:CdSe and Cu+:CdSe NCs shifts, we demonstrate that both also show very similar temperature-dependent PL lifetimes and magneto-luminescence. Electronic-structure calculations further predict that both dopants generate similar localized mid-gap states. Despite these strong similarities, we conclude that these materials possess significantly different electronic structures. Specifically, whereas photogenerated holes in Cu+:CdSe NCs localize primarily in Cu(3d) orbitals, formally oxidizing Cu-i- to Cu', in Agt:CdSe NCs they localize primarily in 4p orbitals of the four neighboring Se2- ligands, and Ag is not oxidized. This difference reflects a shift from "normal" to "inverted" bonding going from Cut to Agt. The spectroscopic similarities are explained by the fact that, in both materials, photogenerated holes are localized primarily within covalent [MSe4] dopant clusters (M = Agt, Cut). These findings reconcile the similar spectroscopies of Agt and Cut-doped semiconductor NCs with the vastly different ionization potentials of their Ag+ and Cu+ dopants.