This text introduces a framework for improving evidence-based policy by moving beyond the simple observation that a program worked in a specific trial. The authors, Nancy Cartwright and Jeremy Hardie, argue that randomized controlled trials (RCTs) only prove a policy functioned "there," which does not guarantee it will work "here" in a different context. To bridge this gap, practitioners must identify the causal principles at play and ensure the necessary support factors are present in the new environment. Using metaphors like "causal cakes," the book provides strategies for determining if a policy can play the same causal role across different settings. Ultimately, the authors advocate for a shift from rigid evidence-ranking schemes toward a more deliberative approach that values local discretion and rigorous thinking. This practical guide aims to help decision-makers estimate their level of confidence in a policy's success by treating evidence as a chain of arguments where every link must be secure.