Ecosystems that exhibit alternative stable states are a prominent challenge for ecological restoration. So far, alternative stable states have been addressed from two different angles: community assembly studies, which focus on species and their interactions, and regime shift studies, which focus on changes in ecosystem states following environmental change. Here, we propose a synthetic perspective that merges the community assembly with the regime shift approach to effectively inform restoration of ecosystems exhibiting alternative stable states. We show that the community assembly and the regime shift approaches have emphasized different aspects of alternative stable states (i.e., coarse vs fine resolutions of the focal state variable, different sets of feedback mechanisms, and small vs large spatial scales), and consequently have different limitations that influence restoration strategies. Using a simple mathematical model, we illustrate that a more explicit consideration of species identity and composition (i.e., the community assembly approach) can improve our ability to understand regime shifts and restore degraded ecosystems. Finally, we highlight two case studies in which such merging can bring novel insights into alternative stable states and ecological restoration. Understanding the relevant aspects of community assembly (biotic interactions and species identity) will lead to more informed decisions that target future restoration and the prediction of regime shifts in response to global environmental change.