In this paper, the local consensus problem of nonlinear time-delay multi-agent systems with switching topologies via distributed saturated impulsive control is discussed and the maximum domain of attraction is well estimated. Specifically, we develop a new estimation approach that is quite distinct from the contractive invariant set to estimate the domain of attraction. Moreover, a novel composite impulsive-instant-dependent Lyapunov function is constructed and an improved convex hull representation with more slack variables is adopted to minimize the conservativeness. And then, some delay-independent sufficient criteria in the form of bilinear matrix inequalities that guarantee local consensus of nonlinear time-delay multi-agent systems are obtained by means of the Lyapunov-Razumikhin technique and proof by contradiction. Through appropriate transformation, some linear matrix inequalities-based optimization problems constrained by the proposed local consensus criteria are formulated, and the corresponding numerical solutions are solved by the Yalmip. Finally, a simulation example is given to confirm the validity and superiority of the proposed theoretical results.