Foraging behaviour governs trophic interactions and can have strong effects on emergent properties of food-web, but the two fields were mainly studied independently in the past. Here I address a largely unresolved question – how does optimal foraging (OF) by consumers influence the food-webs properties? To this end, I compared structure, consumption rate, and dynamical features of empirical food-webs with their OF and randomised counterparts. Structurally, empirical food-webs deviate from OF in-degree prediction to various extents. To link structure to dynamics, I modelled the multi-species population dynamics by using metabolically constrained Lotka-Volterra equations. OF webs, as expected, had the highest total consumption rate, followed by empirical then randomised webs. Larger structural deviations from OF led to larger consumption rate deficiencies. I then evaluated how many species could coexist by analytically solving the equilibrium state, given that structure and interaction strengths are fixed. This food-web feasibility was positively correlated with total consumption rate. I conclude that OF structure contributes to the maximisation of consumption rate and community feasibility. In these two terms empirical food-webs are closer to optimality than randomised ones, indicating that empirically although species forage sub-optimally, the food-web structure it brings about is functionally close to the OF ones.