As practitioners and policymakers talk more and more about personalizing learning, traditional approaches to instruction—which tend to optimize for increasing average student outcomes—must significantly shift. Instead, personalized learning echoes calls for models that serve every student at all levels in each classroom. Blended learning presents one powerful new delivery model to help educators reach this goal. But we must look at blended learning for what is: an instructional delivery mechanism, not the silver-bullet solution to more effective instruction.

Many in the field pose the question, does blended learning work? The truth is, the efficacy of blended learning greatly depends on its specific implementation and the particular problem it is designed to solve. The more pertinent question we should ask is what blended-learning model works, with what software, for which students. As my colleague Michael Horn and I recently wrote about, the paradigm of education research must radically change to properly address these questions. Research and development have long analyzed what works on average for students across the country. Now we need to reframe the research process and evaluate education innovations to understand causality on a deeper level: what works for what students in what circumstances. Only bolstered by precise insights will our schools reap real benefits from online and blended learning that truly deliver on the promise to personalize instruction for each individual student.