Next-generation cosmological surveys will observe larger cosmic volumes than ever before, enabling us to access information on the primordial Universe, as well as on relativistic effects. In a companion paper, we applied a Fisher analysis to forecast the expected precision on fNL and the detectability of the lensing magnification and Doppler contributions to the power spectrum. Here we assess the bias on the best-fit values of fNL and other parameters, from neglecting these light-cone effects. We consider forthcoming 21cm intensity mapping surveys (SKAO) and optical galaxy surveys (DESI and Euclid), both individually and combined together. We conclude that lensing magnification at higher redshifts must be included in the modelling of spectroscopic surveys. If lensing is neglected in the analysis, this produces a bias of more than 1σ – not only on fNL, but also on the standard cosmological parameters.