Whereas previous studies that explored the application of metal–carbon composites as persulfate activators have focused on synergistic performance improvements, the potential advantages or features that can be acquired by integrating metal and carbon constituents that differ in terms of reactivity toward persulfate precursors and their preferred activation routes have been overlooked. With ZIF-67-derived cobalt/N-doped carbon composites (Co@N-C) as the model metal–carbon composite, this study takes a look into a switch in the primary degradative pathway depending on the persulfate precursor used and investigates this kind of composite fabrication as a strategy to overcome the drawbacks of single-component activators. In Co@N-C, Co embedded in the carbon matrix caused radical-induced oxidation in the presence of peroxymonosulfate (PMS) whereas peroxydisulfate (PDS) activation using a carbon framework involved mediated electron transfer. The different nature of the dominant oxidant was confirmed by investigating the quenching effects of alcohols, bromate formation yield, substrate-specificity, electron paramagnetic resonance spectral features, current generation upon sequential organic and persulfate injection, and product distribution. The Co and N-doped carbon serving as separate reactive sites allowed Co@N-C to exploit both PMS and PDS so it could outperform benchmark metal- and carbon-derived materials. Electrochemical measurements linked with X-ray spectroscopic analysis implied that a moderate pyrolysis temperature optimized the Co@N-C activity due to high fractions of graphitic N and Co-N species. Density functional theory calculations reveal that the peroxide bond of PMS is more susceptible to elongation over Co@N-C, thus it is preferentially dissociated to yield sulfate radicals.