Rare events methods are powerful (essentially exact) techniques for overcoming time- and length-scale separations. Several modern rare events methods focus on directly obtaining numerical rates, e.g. forward flux sampling and diffusion maps. Some older rare events approaches for predicting rates can also predict kinetic trends, e.g. dependences on temperature, supersaturation, electrochemical potential, etc. The "classic" theories that share the extraordinary ability to predict trends all share one common theoretical underpinning: they are quasi-equilibrium rate expressions built around a single reaction coordinate with a clear mechanistic interpretation. Motivated by these theories, I will discuss our efforts to discover generalizable reaction coordinates and useful theories for activated processes that as yet remain poorly understood. I will demonstrate our recent progress toward this goal for problems related to ionic crystal growth and for SN2 reactions in aqueous solution.