Governments have 10 years to take back control of their bold agenda. Ceres2030 was an experiment designed to help with the challenge. The project team, employing a complex and rigorous economic model and cutting-edge machine-learning tools, made a partnership with Nature Research that focused on answers to two linked questions: First, what does the published evidence tell us about agricultural interventions that work, in particular to double the incomes of small-scale producers and to improve environmental outcomes for agriculture? And second, what will it cost governments to end hunger, double the incomes of small-scale producers, and protect the climate by 2030? The project focuses on three of the five targets in the second sustainable development goal (SDG 2) and looks at the public spending needed in low- and middle-income countries, including the contribution from donors through official development assistance (ODA).