How do we move from sophisticated models and analytics to evidence that policymakers can actually use? Drawing on a series of projects addressing the U.S. opioid and broader drug overdose crisis since 2018, this talk reflects on our journey from model development to real-world policy engagement. I will use SOURCE, a large national simulation model of the opioid crisis, alongside related projects to illustrate how we have used simulation modeling, data science, and economic evaluation to inform policy decisions. Beyond the technical work, I will discuss our experiences working directly with federal and other decision-makers, adapting to changing policy questions and data limitations, communicating complex model results, and learning where analytics can, and cannot, influence decisions. The talk will share lessons and challenges for INFORMS researchers interested in bringing analytical work into real-world policy settings.