ABSTRACT We present an improved version of the 3D Monte Carlo radiative transfer code possis to model kilonovae from neutron star mergers, wherein nuclear heating rates, thermalization efficiencies, and wavelength-dependent opacities depend on local properties of the ejecta and time. Using an axially symmetric two-component ejecta model, we explore how simplistic assumptions on heating rates, thermalization efficiencies, and opacities often found in the literature affect kilonova spectra and light curves. Specifically, we compute five models: one (FIDUCIAL) with an appropriate treatment of these three quantities, one (SIMPLE-HEAT) with uniform heating rates throughout the ejecta, one (SIMPLE-THERM) with a constant and uniform thermalization efficiency, one (SIMPLE-OPAC) with grey opacities, and one (SIMPLE-ALL) with all these three simplistic assumptions combined. We find that deviations from the FIDUCIAL model are of several (∼1–10) magnitudes and are generally larger for the SIMPLE-OPAC and SIMPLE-ALL compared to the SIMPLE-THERM and SIMPLE-HEAT models. The discrepancies generally increase from a face-on to an edge-on view of the system, from early to late epochs and from infrared to ultraviolet/optical wavelengths. This work indicates that kilonova studies using either of these simplistic assumptions ought to be treated with caution and that appropriate systematic uncertainties ought to be added to kilonova light curves when performing inference on ejecta parameters.