The recombination dynamics in metal-halide perovskite thin films are affected by features that cannot be explained using simplified recombination models that exclude the ability of defects to store charge and re-emit electrons and holes into their respective bands. In this study, the interaction between charge storage on shallow defects and recombination via deep defects is investigated, including exploring whether an abundance of shallow defects that act as dopants only under illumination can improve the performance of photovoltaic devices. At the film level, rate equation models based on the classical Shockley–Read–Hall (SRH) framework explicitly including capture/emission for one shallow and one deep level were used to study the effect of charge storage and re-emission on transient and steady-state photoluminescence as the two most frequently used metrics of recombination dynamics. At the device level, drift–diffusion simulations were conducted to illustrate the effects of defect depths, densities, capture coefficients, transport, and surface recombination velocities on photovoltaic performance. The simulations reveal a regime when deep-defect recombination is strong (deep SRH lifetime <100 ns for the representative parameter set), in which the charge stored on shallow defects leads to a photodoping effect that implies a pronounced asymmetry between electron and hole densities, which suppresses SRH recombination at a given np product. Therefore, the quasi-Fermi level splitting and photoluminescence quantum yield (PLQY) can increase significantly. As a key finding, we identify a regime in which shallow-defect-assisted photodoping is beneficial, yielding a higher quasi-Fermi level splitting and an increase in PLQY by nearly three orders of magnitude compared with the case without shallow-defect charge storage. In a representative device simulation, this translates into an open-circuit voltage increase from 1.06 V to 1.23 V (~170 mV). These findings suggest that shallow-defect-assisted charge storage can be beneficial for performance under certain conditions, pointing to a possible direction for future device optimization.