Halide perovskites solar cells (PSCs) are making true on past promises, having reached power conversion efficiencies (PCEs) of 25.7% and long lifespans (>3000 h). Although stability has become the focus of research efforts, a significant number of researchers are still dedicated to further increasing cell efficiency. To push PCE any higher however, every element of the solar cell must be controlled and optimized. In the context of the recent advancements in halide perovskite doping, we analyse how and why doping can modify the PCE of PSCs. We find that optimal doping levels are highly dependent on carrier mobilities and device architecture, namely whether the hole- or electron-transport layer are on the front-side (illumination-side) of the device. More precisely, there are four regimes defined by carrier mobilities in which different physical processes are more, or less, important causing a change to the optimal doping level. When electron and hole mobilities are comparable, and diffusion lengths are not at least an order of magnitude larger than the perovskite film thickness, devices with the electron-transport layer on the front side (n-i-p) perform better with a p-doped perovskite, whereas devices with the hole-transport layer on the front side (p-i-n) perform better with an n-doped perovskite. The existence of these four regimes is especially pronounced for PSCs due to the very high absorption coefficients and rather low carrier mobilities in halide perovskites. We model the solar cell by employing a drift-diffusion simulation in SCAPS (a Solar Cell Capacitance Simulator) to provide a full rationale for the phenomenon and analyse the conditions under which this effect is significant. The findings presented here are based on the perovskite properties measured by multiple groups and are directed predominantly towards experimentalists working with devices.