Copper nanoparticles are widely used in catalysis and electrocatalysis, and the fundamental understanding of their activity requires reliable methods to assess the number of potentially reactive atoms exposed on the surface. Herein, we provide a molecular understanding of the difference observed in addressing surface site titration using prototypical methods: transmission electron micrscopy (TEM), H2 chemisorption, and N2O titration by a combination of experimental and theoretical study. We show in particular that microscopy does not allow assessing the amount of reactive surface sites, while H2 and N2O chemisorptions can, albeit with slightly different stoichiometries (1 O/2CuS and 1 H2/2.2CuS), which can be rationalized by density functional theory calculations. High-resolution TEM shows that the origin of the observed difference between microscopy and titration methods is due to the strong metal support interaction experienced by small copper nanoparticles with the silica surface.