Earlier, a theoretical model was suggested to discriminate catalytic sites M-n (each consisting of n adjacent atoms M on the metal surface) according to their undercoordination Sigma. It has been shown that the maximum activity of a site M-n (M = Pt, Rh, Ir, Fe, Ru, Re; n = 2, 3, 4) in the catalytic synthesis of ammonia requires the "resonant" Sigma whose major part is inaccessible at perfect planes because of steric restrictions. The current study applies this model to binary alloys and clusters to construct an advanced catalytic site by adjustment of real Sigma to the resonance. The catalytic activity of a site M-n has been estimated by the Bronsted-Evans-Polanyi relation with respect to the formation of NH species. It was found that on alloy surfaces, sites M-3 and M-4 demonstrate synergetic behavior. This suggests that the most active catalyst (Ru or Re) can be improved by its alloying with the least active one (Pt or Rh). In the case of the noble metals, the sites M-3 and M-4 at 4-, 5-, and 11- atomic clusters are similar to 10-10(3) times more active than such sites at perfect planes, whereas the sites of Ru and Re show the opposite behaviour. The model was verified by comparison of the calculated specific catalytic activities of metals, centers Fe-C-7 and Ru-B-5, and single crystals with the published data. The superior activity of a catalytic site is generally enabled by its optimal thermodynamics, which is affected deeply by the first coordination shell. A correlation between local structure, thermodynamics, and activity of a site is likely valid for other catalytic systems.