Co-adsorption is a key initial step in heterogeneous catalysis, which by bringing the reactants together at high coverage on the surface of the catalyst has a clear implication on the catalytic reaction activity. We show that when using Density Functional Theory (DFT) calculations, the choice of the exchange correlation functional can have a qualitative influence on the nature of the obtained most stable coadsorption state. The coadsorption of butadiene and hydrogen, the precursor state for catalytic hydrogenation, is studied on Pt(111) and on the surface alloy Pt2Sn/Pt(111). At typical hydrogenation conditions, the PBE exchange correlation functional gives as most stable situation on both model catalysts a surface fully covered with hydrogen, butadiene remaining in gas phase. This behavior does not agree with available experimental data, and results from an incorrect balance between H and butadiene adsorption energy, mainly by a poor description of dispersion energy for butadiene adsorption. The non-local optPBE-vdW functional provides opposite and correct results, with a co-adsorption of butadiene and hydrogen as most stable situation. The butadiene adsorption energy is strengthened by the description of dispersive forces, hence modifying the nature of the energetic competition between the two adsorbates. The co-asorption energy difference between PBE and optPBE-vdW amounts to 1.04 (resp. 0.7) eV for Pt(111) (resp. Pt2Sn/Pt(111)) on the considered 3x3 unit cell. The computational study of co-adsorption systems from DFT is hence delicate. Errors do not only impact the quantitative adsorption energy of one adsorbate, but they might cumulate over several species, finally providing a qualitatively wrong picture of the optimal coadsorption situation. Going beyond the standard Generalized Gradient Approximation hence appears mandatory for a correct description of catalytic reactivity of unsaturated hydrocarbons.