Implementing Automated Decision-Making (ADM) systems in public administrations raises several tensions: between efficiency and ensuring fair decisions, between public transparency and individual privacy, and between standardization and discretion. To develop legitimate ADM systems that balance these tensions, we need to better understand the phenomenon on the societal, organizational and individual levels. To this end, we conduct a multidisciplinary literature review with the analysis utilizing Coleman's macro-micro model of social action, in which individual attitudes and actions relevant to ADM are related to ADM social structures and outcomes. We develop an ADM framework that captures and conceptualizes these macro-micro relationships and use this framework to identify gaps in research and motivate a research agenda. Our results also reveal a changing decision space between public officials and citizens that, if well investigated, may facilitate the development of citizen-centric ADM solutions and effective human-machine hybrids. We illustrate how the framework and ADM decision space can contribute to research, practice and policy. © 2025 The Authors