Coarse-grained simulation models can provide significant insight into the complex behavior of molecules in the condensed phase. In particular, “bottom-up” coarse-grained models retain chemical specificity by targeting the reproduction of properties from a higher-resolution reference model. However, these models are inherently limited by the molecular representation, set of interaction potentials, and parametrization method. These limitations often result in an inaccurate description of cross-correlations between coarse-grained degrees of freedom, complicating the stabilization of hierarchical structures in soft matter systems. Perhaps more troubling, reduced molecular friction and softer interaction potentials obscure the connection to the true dynamical properties of the system. In this talk, I will present a narrative of our recent progress in characterizing and improving the dynamical properties generated by coarse-grained simulation models. The associated methods that we have developed are largely built upon a Markov-state-modeling framework, which provides a systematic link between microscopic interactions and the emergent long time scale kinetic processes in the system. Our investigations identify structural-kinetic relationships that emerge from restricting certain features of the underlying free-energy landscape, and suggest a concerted approach for a balanced representation of system properties. As model systems, we consider peptide helix-coil transitions and particle jumps in glassy liquids, providing a concrete stepping stone for addressing a wide range of soft matter applications.