Integrated data systems are often used to inform decision making at the top tiers of local or state government. Policymakers use analyses based on large numbers of integrated data system (IDS) records (from which personal identifying information has been removed) to understand the overlap between cases that receive services from different systems and the impact of one system's policies and programs on the outcomes in another. This information helps government officials make smarter decisions about the policies and programs they manage and improve client outcomes. One approach to packaging this information in a way that is readily accessible, easy to interpret and useful for caseworkers is using predictive analytics to develop tools that help caseworkers target scarce resources to those who need them the most and identify the individuals and families who are most at risk. Predictive risk modeling examines large amounts of de-identified data on client characteristics, service receipt and outcomes to determine past patterns in order to predict future outcomes. The human services departments in Allegheny County, Pennsylvania, and Washington state are leaders in sharing IDS data with service providers and developing predictive modeling tools for frontline case managers. This case study describes their innovative approaches, their influence on decision making on the ground and their contribution to improved outcomes for vulnerable children and families.