Reflections from six months as TaRL Africa’s Program Impact Lead
For years, my professional identity was anchored in applied behavioural science. At ideas42, my work revolved around understanding human decision-making, diagnosing psychological bottlenecks, and crafting behavioural solutions to social challenges. It was rewarding, evidence-grounded work.
Yet over time, I found myself increasingly interested in what happens when we zoom out from individual behaviours to the systems and programs in which those behaviours sit. Behavioural science gave me a powerful way of understanding why people do what they do and how we can design better solutions around those insights. At TaRL Africa, I was excited by the opportunity to apply that same way of thinking at a different level: how do you continually improve a program delivered by thousands of people, across thousands of classrooms, within complex education systems?
In February 2026, I made the leap from behavioural science into the role of Program Impact Lead at TaRL Africa.
Too often, “impact” can become an external reporting exercise, managing third-party evaluators, packaging positive metrics for donors, or running experimental projects at the periphery of the core program. At TaRL Africa, the intention was different: to embed diagnostic rigor within the organisation, build a culture of continuous evidence use, and create systems that ensure learning is translated into decisions about the programs we support.
Six months into the role, I have a much better sense of what that looks like in practice, and of what it takes to turn evidence and implementation experience into continuous improvement.
So, what does continuous improvement actually look like, and what drives it?
The evidence for TaRL’s impact is strong, though much of it still comes from India. Since TaRL Africa’s inception, we and our partners have been working to understand how the approach translates across different African contexts. Pre-post data from millions of children suggest 15–40 percentage-point improvements in paragraph reading and simple subtraction over the course of a year. Recent RCTs in Côte d’Ivoire, Madagascar and Zambia also find positive, though varying, effects. Alongside the formal evidence, TaRL Africa now has more than a decade of experience working with partners to implement and scale the approach.
Reflecting on this evidence and experience, there has been real progress in what we know and what TaRL has achieved, but there is also a deep commitment within the organisation to keep learning, improving, and doing more for the learners we aim to serve.
For an organization focused on helping learners gain foundational skills, the challenge is not simply to ask, “Does TaRL work?” It is to keep asking where it is working, where it is falling short, why those differences exist, and what we can do about them. That means looking beyond learner outcomes or classroom practice alone, and trying to understand how different parts of the system and program shape what happens in practice.
As I have gotten deeper into the work and engaged more closely with other frameworks, such as the What Works Hub for Global Education Core Components of TaRL, I have found it useful to think about impact as the result of several things coming together: the conditions created by the education system, the design of the program, and how well the approach is implemented in classrooms.
Drivers of Impact Framework
- System conditions are the broader education-system factors that enable or constrain TaRL implementation, including the extent to which TaRL is authorised, recognised, supported and integrated within the system, and whether system actors have the buy-in, capacity, structures and resources needed to implement it.
- Program design is the quality and coherence of the TaRL model: how we assess and group learners, sequence instruction, use activities and materials, use assessment data to target children, classes, schools, districts and provinces with tailored support, and prepare and support teachers.
- Classroom implementation is where these elements come together in practice: the consistency and quality with which teachers are able to translate the TaRL model and the support they receive into effective instruction.
These elements are closely connected. System conditions shape whether and how TaRL can be implemented, program design shapes what teachers are asked and supported to do, and classroom implementation determines how well the approach translates into learning. Underlying all of this are the mindsets of the people involved, whether they understand the learning problem, care about it, see improving learning as their responsibility, and believe all children can learn.
TaRL also does not operate in the classroom in isolation. Effective implementation requires engaging with and designing for multiple layers of the wider system, from system-level structures and policies, to district and school-level support, to classroom practice. This way of thinking has shaped how I approach the role: identifying where the program can be strengthened and turning those insights into action.
Getting specific about what good implementation looks like
One of the first things I wanted to understand was what effective TaRL implementation actually looks like in the classroom. It is easy to ask whether a program is “being implemented well.” It is much harder, and much more useful, to define what that actually means in terms of what teachers do.
My behavioural science background came in handy here. We unpacked the TaRL model into 16 key teacher behaviours that we believe are important for effective implementation. These range from behaviours that happen once or occasionally, such as setting up learning groups correctly based on assessment data, to behaviours that teachers need to practise consistently, such as selecting activities that are appropriate for learners’ current level.
Defining these behaviours gives us a concrete picture of what we are asking teachers to do and, importantly, a starting point for understanding what needs to be in place for those behaviours to happen consistently. And this is where the broader drivers-of-impact framework becomes useful. If a behaviour is not happening, we can look beyond the teacher to understand what might be shaping it, whether the issue lies in the program design, the support and conditions around the teacher, or the wider system.
This shifts the question from simply “Why aren’t teachers doing this?” to “What is driving this behaviour, and what needs to change for it to happen consistently?” That might mean changing something in our own programs, changing how we work with and support government partners and other partners, or working with government and others to address barriers that sit beyond TaRL.
Strengthening instructional design
Having defined what effective implementation looks like and begun to understand the factors that shape it, the next question was where we should work with partners to strengthen the instructional model. This is one driver of impact where we can have relatively direct influence: working with government and implementation partners, we can identify areas for improvement, adapt the instructional model, activities, guidance and tools, and test and learn from those changes together.
Across the Zambia and Côte d’Ivoire RCTs, as well as our internal data, we have seen that while many learners are developing basic decoding skills, fewer are progressing to reading paragraphs and stories fluently. This has prompted us to look more closely at the progression within the literacy model and ask what we could strengthen it to better support learners along that journey.
This has led us to pilot a number of adaptations to the literacy model. We are piloting a systematic phonics scope and sequence, a planned progression for teaching the relationships between letters and sounds, moving from simpler to more complex skills, to explore whether it gives learners a clearer pathway to become readers. We are also leveraging decodable texts aligned to the phonics progression, giving learners more opportunities to apply what they are learning to connected text.
We are introducing additional sentence-reading and fluency activities, alongside revised learner grouping that, where contextually possible, treats Word-level learners as a distinct group with activities designed around their specific needs.
We are also providing clearer guidance on how teachers select and sequence activities based on the skills they are trying to develop, which aims to make it easier to provide more targeted, skill-based instruction.
Finally, we have looked at the tools teachers use to deliver the model. In collaboration with ideas42 we redesigned the teacher manuals through a behavioural science lens, with a more consistent activity layout and clearer activity grids to make them easier to navigate in the flow of teaching and to support teachers in deciding what to teach.
These are not changes we can assume will work simply because they make sense on paper. They are hypotheses that we need to test. Our aim is to use evidence and implementation experience to identify where the model can be strengthened, make targeted changes, and learn from what happens next. And we may find that some of the changes we initially hypothesise are not the ones that meaningfully shift outcomes. The process may instead point us to other drivers in the drivers of impact framework that need to be strengthened or changed. This is precisely why we are taking an iterative, evidence-informed approach: to learn what actually makes the biggest difference to impact.
Strengthening teacher support
Of course, strengthening the instructional design is only part of the equation. Teachers also need to be prepared and supported to translate that model into practice. This is an area where we have been working for some time, and where we have found that there are no simple fixes.
For mentoring, we’ve tried different approaches, from remote mentoring to peer mentoring, and are now working to strengthen school-based mentoring and support. So far, this work has not given us a clear answer about what works best. For training, we’ve similarly explored different models, from longer and shorter training to staggered approaches, and we are still learning what approaches are most effective in different contexts.
What this has reinforced for us is that there is unlikely to be one standard way to train or mentor teachers effectively. What works depends on the context, the people providing support, and the conditions in which teachers are working.
Training
At the same time, our experience and external research are helping us build a clearer picture of the principles that should underpin effective in-service training: treating it as a continuous process rather than a one-off training event; applying principles of adult learning; focusing on a small number of clear instructional practices; and creating enough opportunities for modelling, practice, feedback and reflection. The challenge is figuring out how to apply these principles in ways that work in different contexts and are feasible and sustainable at scale.
We are beginning to translate these principles into changes across the training journey. We’ve restructured training around the key teaching and support behaviours required for effective TaRL implementation, with more opportunities for teachers to practise and reflect rather than relying primarily on explaining what good implementation looks like. We’ve also strengthened guidance on practical aspects of implementation we know teachers struggle with, including session planning and activity selection, continuous learner progression and managing large classes.
We have also developed a new Trainer’s Manual to strengthen Master Trainer preparation and provide more structured guidance for those delivering the training. We’ve introduced short quizzes throughout training to check understanding and identify where participants need further clarification or support. We are also delivering Master Trainer training without slides, so that the training Master Trainers receive reflects the conditions in which they will be expected to train teachers.
Teacher support
But training is only the beginning. What happens after teachers return to their schools matters just as much. Here too, we have learned that there is no one-size-fits-all approach. Ongoing support needs to work within the structures and roles that already exist in each context, while ensuring that teachers have the conditions and support they need to implement TaRL well.
This has led us to think more deliberately about what we mean by teacher support. Support is often understood primarily as mentoring, but effective TaRL implementation requires much more than mentoring alone. Teachers need the time, materials, structures, expectations and leadership support that enable them to implement the approach well, alongside ongoing support to strengthen their instructional practice.
We are therefore strengthening two complementary forms of support:
- Implementation support: Different actors across the system help create the conditions for TaRL to happen well, including ensuring TaRL time and materials are in place, coordinating support, clarifying expectations, reviewing implementation and addressing challenges.
- Instructional mentoring: Mentors provide teachers with ongoing, practical support to strengthen their classroom practice through observation, feedback and reflection.
To support these two distinct roles, we have developed a clearer matrix of roles and responsibilities, alongside practical tools such as an annual planning checklist for school leaders to help them set up the enabling conditions for TaRL, a simple classroom observation tool to guide mentors in observing and discussing practice, an assessment verification tool to help check assessment results, and a mentoring FAQ covering issues such as frequency, feedback and building positive mentoring relationships.
Teacher training in Zambia
Knowing what to change and what to protect
There is another side to continuous improvement that I have been thinking about just as much: knowing what not to change.
TaRL has a simple but powerful premise: start where the child is and teach at that level. As we adapt the model, it is important that we don’t lose sight of what makes that approach distinctive in the first place.
For me, this means that every change needs to be considered not only in terms of whether it might improve learning outcomes or make implementation easier, but also in terms of what it preserves or risks losing, for the learner and teacher, and for the sustainability of the program and system.
Does this change help us better meet learners where they are? Does it make it easier for teachers to respond to learners’ actual needs? Does it preserve the sense of progress and possibility that comes from teaching children at a level where they can succeed? Does it hinder government ownership and prospects for durability?
These questions are particularly important when working at scale. A model that is implemented across thousands of classrooms will inevitably need to adapt to different contexts, systems and constraints. But adaptation should not mean losing the principles that sit at the heart of the model.
This is perhaps one of the tensions I have come to appreciate most in the role: continuous improvement requires both a willingness to change and a clear sense of what should remain constant.
We need to be willing to question the model, respond to evidence and learn from implementation. But we also need to stay anchored to the reason we are making those changes in the first place: helping children who have fallen behind develop the foundational skills, confidence and belief that they can learn.
That balance, evolving the model while protecting its essence, is something I expect we will keep navigating as TaRL Africa continues to grow.
Six months in
Six months into the role, I have a much clearer sense of what an Impact Lead can bring to an organisation. It is about bridging insight into action, connecting what an organisation is learning from evidence and implementation experience with the decisions it makes about its program. It means translating evidence and experience into practical adaptations, supporting implementation, and learning from what happens next. It also means moving between the learner, the teacher, the school and the wider system, and understanding how each shapes what ultimately happens in the classroom.
In some ways, this feels like a long way from where I started in behavioural science. But in other ways, it feels like a return to it.
I still spend a lot of time thinking about behaviour: what teachers do, what learners experience, and what makes it easier or harder for teachers, school leaders and other actors across the system to do what we hope they will do. The difference is that I’m now thinking about behaviour not in isolation, but as part of a much larger system. I’m thinking about how the actions of different actors, the design of the program, the conditions around it, and people’s mindsets interact to shape what ultimately happens in the
Perhaps that is what “from behavioural change to impact, and back” means for me. I haven’t left behavioural science behind. I’ve simply zoomed out, from changing individual behaviours to understanding the systems and programs that shape them, and then back in again, with a broader view of what it takes to turn good ideas into better outcomes at scale.
And finally, a note of gratitude. While I’ve written this from the perspective of my own experience in the role, this work is very much a team effort. I’ve been fortunate to work with colleagues who care deeply about improving learning outcomes for millions of learners and much of what I described here has been shaped by the work, ideas and perspectives of many.
Kaitlin Alsofrom,Olubunmi Adefisayo,Ashleigh Morrell, Pumza Ndamase, Anjali Shandilya,Daniele Ressler, Hafsatu Hamza, Amos DEMBELE, Emily Joof, Maxwel Ajuoga, Ben Simuyandi, Emmanuel Omurunga, Umar Isyaku, Sylvester Mchihi, Sylvia Njuma, Titus Syengo, Anna Carmen Murru, Alia Karim, Arja Dayal, Bhekilizwe Maphosa, Raphael Walubengo, PMP®, Muideen Olawale Jimoh, and Catherine Botros-Odingo just to name a few…
