Binary metal oxides have received sustained interest as anode materials due to their desirable capacities, exceeding theoretical values particularly in the first discharge. Although they have received increasing attention in recent years, topical debates persist regarding not only their lithiation mechanisms but also the origin of additional capacity. Aiming to resolve these disagreements, we perform a systematic study of a series of iron and manganese oxides to investigate their phase behavior during first discharge. Using a combination of in operando pair distribution function measurements and our recently developed Metropolis non-negative matrix factorization approach to address the analytical challenges concerning materials’ nanoscopic nature and phase heterogeneity, here we report unexpected observation of non-equilibrium FeO x solid-solution phases and pulverization of MnO. These processes are correlated with the extra capacities observed at different depths of discharge, pointing to a metal-dependent nature of this additional capacity and demonstrating the advantage of our approach with promising prospects for diverse applications. The extra capacity may have a dominant source depending on specific metal species bcc -FeO x intermediate lithiate via oxygen extraction from the Fe sublattice Local structure characterization using pair distribution function Robust phase deconvolution using Metropolis non-negative matrix factorization Binary metal oxide anodes in lithium-ion batteries show extra capacities at their initial discharge. In this article, Hua et al. perform a systematic study of iron and manganese oxides’ lithiation phase behaviors and report unexpected observations of bcc -FeO x solid solution and MnO pulverization, processes correlated with the extra capacities seen at different depths of discharge.