In the business of making policies, decisions are based on experience and guided by political concerns. However, in the business of delivering policies, the machinery of government is often taxed by delays and inefficiencies, and constrained by insufficient resources, management tools, and just-in-time information. The result is that governments operate well below the efficiency frontier. For most of our history, research has been disconnected from policy and has moved slowly to build knowledge relevant to designing policies. The authors introduce some of the principles that govern this young institution in the chapters that follow, each designed to exemplify the value of doing better research for doing better development. In these chapters, they present four overarching ideas that they have worked into development practice. Overall, Development Impact Evaluation (DIME's) approach is to inform the path of development through a capacities-based and iterative process of evidence-informed adaptive policy change. To do so, DIME has developed and implemented a model of co-production with agencies on the ground that transfers capacity and know-how to partners, enables them to make mid-course corrections and motivates the scale-up of more successful policy instruments to achieve policy outcomes and optimize development impact. Finally, DIME invests in public goods to improve the quality and reproducibility of development research around the world.