In response to overwhelming evidence of the learning crisis across the globe, parents, governments, and donors are increasingly asking: “What can we do to improve student achievement?” In particular, they are eager for solutions that will work within existing, resource-constrained school systems.

In 2006, when the Hewlett Foundation started the Quality Education in Developing Countries initiative, one of the initiative’s goals was to help answer precisely this question. From 2007 to 2013, combining resources with co-funding from the Bill & Melinda Gates Foundation, the Foundation supported eleven school-level approaches to improving early learning, accompanied by ten rigorous evaluations. The grants spanned India and five countries in Sub-Saharan Africa: Kenya, Uganda, Mali, Senegal, and Ghana. Most organizations focused on improving instructional practice, tackling head-on the fact that teachers are insufficiently prepared to teach reading and math in the early primary grades.

The studies funded by the Foundation constitute a significant contribution to existing evaluation evidence on how to improve student learning. Recent meta-analyses catalogue fewer than 80 randomized studies in the developing world that have examined learning as an outcome variable (McEwan 2013, Krishnaratne et al. 2013). Given the paucity of existing studies, it is still early days for building evidence on what works to improve learning.

The remainder of this paper provides a synthesis of what we know about how to improve learning outcomes. Three areas seem to be critical: (1) improved instruction; (2) strong teacher training and in-school mentoring; and (3) community engagement in learning. The paper concludes with recommendations for carrying this work forward, including ways the results could help shape the future research agenda.