Cosmological surveys planned for the current decade will provide us with unparalleled observations of the distribution of galaxies on cosmic scales, by means of which we can probe the underlying large-scale structure (LSS) of the Universe. This will allow us to test the concordance cosmological model and its extensions. However, precision pushes us to high levels of accuracy in the theoretical modelling of the LSS observables, in order not to introduce biases in the estimation of cosmological parameters. In particular, effects such as redshift-space distortions (RSD) can become relevant in the computation of harmonic-space power spectra even for the clustering of the photometrically selected galaxies, as it has been previously shown in literature studies. In this work, we investigate the contribution of linear RSD, as formulated in the Limber approximation by arXiv:1902.07226, in forecast cosmological analyses with the photometric galaxy sample of the Euclid survey, in order to assess their impact and quantify the bias on the measurement of cosmological parameters that neglecting such an effect would cause. We perform this task by producing mock power spectra for photometric galaxy clustering and weak lensing, as expected to be obtained from the Euclid survey. We then use a Markov chain Monte Carlo approach to obtain the posterior distributions of cosmological parameters from such simulated observations. We find that neglecting the linear RSD leads to significant biases both when using galaxy correlations alone and when these are combined with cosmic shear, in the so-called 3$\times$2pt approach. Such biases can be as large as $5\,\sigma$-equivalent when assuming an underlying $\Lambda$CDM cosmology. When extending the cosmological model to include the equation-of-state parameters of dark energy, we find that the extension parameters can be shifted by more than $1\,\sigma$.