Teaching at the Right Level is a ‘smart buy’ for learning, but we need to rethink how we support teachers

“Teaching at the Right Level” (TaRL) is one of the best-evidenced ideas in global education. A synthesis of 14 evaluations across seven countries found that, when children actually take up the program, it delivers average learning gains of 0.42 standard deviations, a significant enough effect that makes TaRL stand out as a cost-effective option for governments trying to solve the learning crisis on a tight budget. But that same synthesis found that effect sizes varied widely, from 0.08 to 0.70 standard deviations, and one of the two factors that explained most of that variation was who delivered the program. Volunteer-led implementations consistently outperform teacher-led ones. Since national governments scale programs through their teaching workforce, this becomes the central design question for anyone trying to take TaRL from a small pilot to the system level.  

A recent two-year study in Côte d’Ivoire offers a close-up view of why that gap exists and what it would take to close it. Côte d’Ivoire has been testing and scaling a government-led version of TaRL, the Programme d’Enseignement Ciblé (PEC), in the country’s cocoa-growing regions since 2018 and nationally as part of the Programme National d’Amélioration des Premiers Apprentissages Scolaires (National Program for Improving Early School Learning) or PNAPAS reform. This piece reports findings from that randomized controlled trial measuring the impact of PEC in Côte d’Ivoire and maps them against Brookings’ 2021 scaling case study of the same program to draw recommendations for scaling foundational learning reforms.  

What the study found 

Wolf et al. (2026) find that PEC modestly improved children’s reading and math learning levels: a 6-percentage-point drop in the share of children stuck in the bottom literacy group and a 10-percentage-point drop in the bottom numeracy group. Although these effects were statistically significant, they were less practically meaningful. Importantly, when looking at more detailed reading and math assessments designed to capture functional skill rather than coarse group placement, the effects were only about half the size and only marginally significant (meaning the evidence wasn’t strong enough for us to be confident that the effect was real). Two years in, most children (63%) still weren’t reading even at a basic word level.  

This broadly matches what teacher-led, government-scale TaRL programs have produced elsewhere. On the why, the data pointed squarely at teacher support as the main mediator of impact: 

• Teacher knowledge decayed fast. A one-week training initially raised teachers’ scores on a PEC knowledge test, but that knowledge dropped significantly by the end of the school year. It never recovered, even after a refresher training the following year and despite ongoing access to mentors and a chatbot support tool built for this purpose. Critically, it was knowledge measured right after the initial training, not later, that predicted student learning outcomes at the end of the study. In other words, the front-loaded training mattered most for student learning, and its effects faded before the two years were up. 

• Training is expensive, especially at scale. The Brookings Real-Time Scaling Lab report found that training teachers and mentors accounted for roughly 80% of total PEC implementation cost and that the training as delivered at the time would not be affordable for the government at national scale. In addition to not being sustainable at scale, this recent study confirms that the high costs of training are not buying durable knowledge either. The study’s findings suggest that a large share of that training investment may depreciate within a year. 

• Burnout nearly halved the program’s impact. Treatment effects on children’s learning levels were roughly twice as large in classrooms where teachers reported low burnout at baseline. Unsurprisingly, teachers arriving already depleted had less capacity to take up a demanding new instructional approach. The Brookings analysis flagged that Côte d’Ivoire “lacks sufficient education personnel” to deliver the program with quality at scale, a concern that the new findings on teacher burnout now bear out.  

• Teachers experienced the program as an addition to their job, not a change in how they do it. In a qualitative study, teachers described real enthusiasm for what PEC did for their students. But they also described PEC as competing for time and energy with their existing obligations, rather than replacing or streamlining them. As one teacher put it, government-led reform was supposed to make the job easier; instead, “the two compete.” 

Why this matters for funders, implementers, and ministries, not just researchers 

PEC offers an example of a pattern seen repeatedly: an intervention with strong evidence gets adopted at scale and then underperforms its own pilot data—not because the pedagogy was wrong, but because the system conditions for delivering it well were treated as an implementation detail rather than a core design problem. This piece offers specific, fundable fixes related to scaling quality education reforms: 

• Treat refresher training and coaching as recurring line items, not one-off costs. Knowledge decays within a year, so a single preservice training week is not enough for a multi-year rollout. As noted above, a significant portion of teachers in Côte d’Ivoire left the training without acquiring basic program knowledge, suggesting the initial training needs improvement. But simply adding more intensive training is not an easy fix; scaling PEC nationally as-is over five years would cost more than 15 times the education budget allocated to in-service teacher training for that period. A more affordable and sustainable long-term option is integrating PEC into preservice teacher education to reduce reliance on in-service sessions. 

• Invest in low-cost, ongoing channels alongside in-person mentoring. Program knowledge is necessary but not sufficient to meet the complex challenges of a real classroom; ongoing support is also important. But rural distances and advisor shortages place real constraints on how much continuous assistance in-person mentoring can provide (a constraint the Brookings report also identified). To help bridge this gap, the study tested a chatbot-based support tool (DIA) that was popular with teachers and gave program administrators visibility into where knowledge gaps persisted. The Ivorian government has since scaled PEC nationally using a hybrid digital training model, aiming to use technology to help balance cost, efficacy, and reach. However, this approach has moved away from the fully in-person model used in most other TaRL implementations, and we don’t yet have evidence that this trade-off maintains training quality. So, while looking for affordable ways to provide ongoing teacher support is critical, understanding what is lost or gained in that transition remains an open and important question. 

• Design the program to reduce workload, not add to it. Teachers were far more positive about PEC’s effect on students than on their own working lives. Aligning targeted-instruction time with existing curriculum planning would likely improve fidelity and reduce the burnout that the data show is suppressing impact. This is critical for teacher well-being and key to program effectiveness. Research has indeed shown that teachers play a huge role in student learning outcomes.  

If we want government-led targeted instruction to close the gap in student learning, scaling and sustaining quality teacher support is critical.