Although grafting polymers onto surfaces is widely suggested for designing smart systems, optimizing the performance of such systems is not simple. In this article, we investigate an azo-polymer-based smart surface using the single-chain-in-mean-field theory. Through the numerical simulations, we study the adhesion/erasion transition of the system and show that the performance of the smart surface can be characterized by the difference between the effective nanoparticle-surface interactions in the UV-on and UV-off states. Further exploring the optimization of the smart surface, we find that the distribution function of the receptor can have typical bimodal characteristics, which is crucial for optimizing the position of the azo-bond along the azo-polymer, f. Moreover, the presence of the homopolymer is also essential for the optimal performance of the smart surface, and we build a reference map for the good combinations of f and the homopolymer design f(homo).