Depression is the leading cause of disability worldwide — yet knowledge of actionable strategies that could mitigate depression risk at the population level remains limited. Using genomic and phenotypic data from over 100,000 individuals, the authors demonstrate a two-stage approach to systematically identify and validate a wide range of potential modifiable factors for depression. This approach consisted of an exposure-wide association scan followed by two-sample Mendelian randomization analyses. An array of promising targets for depression were identified using this approach: most robustly, social support-seeking and interactions, daytime sleep patterns, and screen-based media use — as well as potential physical activity and dietary factors. A data-driven, genetically informed approach could help prioritize intervention targets for follow-up studies including clinical trials for depression prevention.