We study the impact of the often neglected lensing contribution to galaxy number counts on the Eg statistics which is used to constrain deviations from GR. This contribution affects both the galaxy-galaxy and the convergence-galaxy spectra, while it is larger for the latter. At higher redshifts probed by upcoming surveys, neglecting this term induces a substantial error in the spectra and therefore on the Eg statistics which is constructed from the combination of the two. Moreover, including it, renders the Eg statistics scale and bias-dependent and hence puts into question its very objective.
