Whether it be in the context of antibiotic treatments or exponential growth in constant environments, non-genetic variability has shown to have significant effects on population dynamics. I will first present a coarse-grained model for cell growth, inspired by the Langevin equation, which incorporates both biomass growth rate and generation time fluctuations. Building on it, we will connect single-cell variability to the population growth, showing that in contrast to the dogma growth-rate variability may lower the population growth. Analogous results apply to the case where the variability arises from the asymmetric partitioning of a cellular resource, where we find a phase transition between a regime where variability is beneficial to one where it is detrimental. We will also show that a population's growth rate can be inferred from studying a single lineage, with intriguing relations to large deviation theory underlying a non-monotonic convergence of the estimate on lineage length.