ABSTRACT We analyse synthetic 12CO, 13CO, and [C ii] emission maps of molecular cloud (MC) simulations from the SILCC-Zoom project. We present radiation, magnetohydrodynamic zoom-in simulations of individual clouds, both with and without radiative stellar feedback, forming in a turbulent multiphase interstellar medium following on-the-fly the evolution of e.g. H2, CO, and C+. We introduce a novel post-processing routine based on cloudy which accounts for higher ionization states of carbon due to stellar radiation in H ii regions. Synthetic emission maps of [C ii] in and around feedback bubbles show that the bubbles are largely devoid of [C ii], as recently found in observations, which we attribute to the further ionization of C+ into C2+. For both 12CO and 13CO, the cloud-averaged luminosity ratio, $L_\rm {CO}/L_\rm {[C\, \small {II}]}$, can neither be used as a reliable measure of the H2 mass fraction nor of the evolutionary stage of the clouds. We note a relation between the $I_\rm {CO}/I_\rm {[C\, \small {II}]}$ intensity ratio and the H2 mass fraction for individual pixels of our synthetic maps. The scatter, however, is too large to reliably infer the H2 mass fraction. Finally, the assumption of chemical equilibrium overestimates H2 and CO masses by up to 150 and 50 per cent, respectively, and $L_\rm {CO}$ by up to 60 per cent. The masses of H and C+ would be underestimated by 65 and 30 per cent, respectively, and $L_\rm {[C\, \small {II}]}$ by up to 35 per cent. Hence, the assumption of chemical equilibrium in MC simulations introduces intrinsic errors of a factor of 2 in chemical abundances, luminosities, and luminosity ratios.