In East Africa, there is great effort directed toward ensuring that there is learning and value for money invested in universal education policies initiated over the past decade. Kenya and Uganda are two countries that typify this effort. The effort includes the work of research organisations such as Uwezo, which assess learning levels; RTI, which assesses language and early grade reading; and the work of African Population and Health Research Center (APHRC), which looks at schooling patterns of different households. All these studies paint a disappointing picture both for the universal access Education for All policy and the large sums of money that have been devoted to achieve it. The verdict is that too many pupils in schools are not learning and too many poor ones are excluded from universal public access. Uwezo reports that 70% of pupils in Grade 3 cannot do Grade 2 work, and 9% of those completing Primary 8 in Kenya cannot do Grade 2 work. Answers are not easy to find, so borrowing what has worked elsewhere has been attempted. This article presents one effort toward finding what can work to improve learning for pupils in early grades. It is based on the idea of “Reading to Learn” implemented elsewhere and attempted in East Africa by Aga Khan Foundation and independently evaluated by APHRC using randomization methods. Lessons presented highlight the role and complexities of randomization in addressing the educational challenges in East Africa.