Data-driven decision making (DDDM) has been a catchphrase in education for the past decade. Everyone is for it, and who could object? Most agree that decisions from the classroom to the central district office and the state education agency should be informed by good data. In practice, however, it is all too easy for data to leave educators and policymakers unmoved--or, worse yet, to drown them in extraneous information--rather than drive decisions that will improve classroom instruction, school performance, and student achievement. This panel proposes to examine DDDM from a variety of perspectives. First, presenters will discuss a framework that provides a comprehensive picture of the DDDM process in education, starting with a high-level, generalized theory of action--a causal chain--for how DDDM can lead to improved student achievement, and the supports and incentives needed to make effective data use possible. The framework also maps the process of DDDM at different levels of the education system, from classroom to state superintendent’s office, depicting the types of decisions that might be informed by data, the types of data needed to inform different decisions, and the importance of determining that the data are both relevant and diagnostic. Next, a school district administrator and a Strategic Data Project (SDP) Data Fellowship Alumnus will comment on lessons learned from a practitioner’s viewpoint on how school districts can augment the use of data and evidence in improving strategy and decision-making. Since 2008, SDP at Harvard University has partnered with school districts, charter school networks, state education agencies, and nonprofit organizations to bring high-quality research methods and data analysis to bear on strategic management and policy decisions. One of SDP’s strategies is to build a network of top-notch data strategists who serve as fellows for two years with a partner agency. Third, the founder of Alumni Revolution, an organization that provides support to first-generation college students with the aim of helping them toward graduation, will describe the critical role of data in its efforts, notably including an algorithm that identifies the best mentor match for each mentee participating in the program. Finally, researchers will describe school-level data use from a study of personalized learning (PL) environments in schools. PL models tailor instruction to individual student needs so an important component of most PL models is a competency-based system that involves assessing students’ initial performance, placing them in appropriate instructional environments, and using data on performance to determine whether students advance to the next grade or level. Effective implementation requires educators to have access to high-quality data on student progress and to understand how to use those data appropriately for instructional decision making.