Implementation Still Matters Whitepaper

In 2010, the American Institutes for Research conducted a descriptive study of Read 180^® across five urban middle schools to understand why the same reading intervention produced markedly different results from one district to the next. The researchers found that the quality of implementation, not the program, was responsible for those differing results. The implementation and results were dependent on the conditions districts and schools created, or failed to create, around the program. The researchers organized these conditions into three phases of implementation: initiating, developing, and sustaining (Salinger et al., 2010).

Sixteen years later, the three phases of implementation still resonate, but the research base has grown substantially, and the stakes have risen. The post-pandemic recovery period has demonstrated this at a national scale. Districts invested billions in strong, evidence-based interventions, yet most recovery programs experienced shortfalls of implementation and reached fewer students, at lower dosage, than planned (Carbonari et al., 2024). As districts face increasing scrutiny of education technology spending and the rapid adoption of AI-enabled tools, the lesson is clearer than ever: the difference between an instructional program that improves student outcomes and one that disappoints is in how well it is implemented. This brief revisits each of the original lessons learned, updates the lesson with what the research now demonstrates, and looks ahead to what effective implementation will require in the era of AI.