Microbial communities often face external perturbations that can induce lasting changes in their composition and functions. Responses to such perturbations can drive community functions, including host health, which urges for predictive understanding of community dynamics following perturbations. In this talk, I will present our work addressing how perturbations can induce transitions between alternative stable states of a simple experimental community. For the case of antibiotic perturbations, I will address how bacterial growth dynamics can overcome differences in antibiotic susceptibility and determine the community outcome of antibiotic exposure. Combining theory and experiments, we demonstrated that differences in the growth rates of community members can determine the resilience of stable community states against several antibiotics, a result that holds with relative independence of differences in antibiotic susceptibility. In our path to predict the dynamics of more complex communities, we then faced a classical question in ecology: how does community complexity impact stability and dynamics? We addressed this question by assembling experimental communities under a large range of conditions, controlling both identity and number of community members, as well as different environmental parameters. These experiments constitute the first empirical demonstration of complexity-driven transitions between three generic phases of community dynamics predicted by ecological theory.