We present a comparison between accurate ab initio calculations and a high-quality experimental data set (1986-2002) of electric-field gradients of Cd at different sites on Ni, Cu, Pd, and Ag surfaces. Experiments found a systematic rule to assign surface sites on (100) and (111) surfaces based on the main component of the electric-field gradient, a rule that does not work for (110) surfaces. Our calculations show that this particular rule is a manifestation of a more general underlying systematic behavior. When looked upon from this point of view, (100), (111) and (110) surfaces behave in precisely the same way. The physical mechanism behind the systematics of the EFG for other 5sp impurities (Cd-Ba) at different fcc surfaces sites is revealed, showing in a natural way why the first half of the 5p elements shows a coordination dependence that is opposite with respect to the second half.