The electrochemical reduction of CO₂ is a promising technology to reach a carbon-neutral economy. However, among other challenges, the design of active and selective catalysts still limits such advances. Herein, we explored the atomistic engineering of the catalyst substrate as a strategy to tune Cu catalysts for CO₂ reduction towards different C₁ products by using generalized coordination numbers, GCN, as a structural descriptor of catalytic properties. We observed bifurcations on reaction mechanisms and significant changes in the predicted onset potentials, U_onset, towards CO, HCOOH, and CH₄ as a function of the GCN values of the adsorption sites. For each product, we observed volcano plots of U_onset as a function of GCN that indicate activity peaks. We also compared the evolution of U_onset for different products to show that, regardless of the small U_onset values for CH₄ on low-coordination sites, such adsorption sites should be more selective towards CO. For Cu nanoparticles, the analysis showed that decreasing the nanoparticle size could lead to more active catalysts that should be highly selective towards CO, which qualitatively agrees with experimental results.