Development cooperation organizations invest heavily in results reporting systems to identify ‘what works’. However, new research reveals that much learning happens at earlier stages of the project cycle. If reporting is not a major driver of learning or decision-making, then the extensive – and costly – systems currently in place may need to be reconsidered, and more emphasis should be placed on project planning.
International development cooperation is under growing pressure. Shrinking budgets, rising political scrutiny, and persistent claims of ineffectiveness have fueled debate over whether it should continue at all. A common critique is that International Development Organizations (IDOs) fail to deliver results – or deliver results but fail to learn from them. In fact, entire systems have been built to produce and validate performance data that often appears to be only loosely connected to actual decision-making.
How much and what type of information do IDOs need before they can learn and adapt their operations? Previous research offers a wide range of explanations for this failure to learn, including high uncertainty, temporal differences, and long distances between donors and recipients; lack of reliable data in development settings; a focus on upwards accountability that ‘crowds out’ learning: that politics limits evidence-based policymaking, among others.
However, in new research published in World Development, we suggest a different angle and begin by taking a closer look at when exactly learning occurs. We argue that development organizations ‘de-couple’ learning from results reporting. So, rather than assuming that results reporting will drive learning, it would make more sense to explicitly link learning processes to decision-making points – especially during planning and foresight activities.
For example, when staff engage in envisioning real solutions to problems, which is often a requirement when planning development projects, they are more likely to seek out relevant evidence and engage in meaningful dialogue. This echoes research suggesting that the incompleteness of information during the production of results management tools like the ‘Balanced Scorecard’ at the early stages of project development motivates learning.
There is also a case for streamlining reporting requirements. Results reporting is often considered a boring accountability exercise disconnected from specific decision-making points. In many cases, reporting diverts resources away from implementing activities. If reporting is not a major driver of learning or decision-making, then the extensive – and costly – systems currently in place may need to be reconsidered.
What are the key reporting and impact questions for development researchers today?
Could reporting be made leaner? Could resources instead be redirected toward foresight and sense-making? These questions are particularly urgent in an era of tightening budgets.
Arguably, it is time to rethink the dominant results management framework (which has been in place since the 1990s). Results-Based Management (RBM), used by virtually all aid organizations – from bilateral organizations such as Norad or the Foreign, Commonwealth and Development Office (FCDO) to multilateral institutions such as the World Bank and the United Nations – has long guided the way development work is planned and assessed.
Its promise is straightforward: by systematically collecting and analyzing performance data, organizations can improve decision-making, increase efficiency, and ensure that public funds are used effectively. In theory, RBM creates a learning loop in which organizations plan interventions, implement them, measure results, and then use those results to inform future decisions. But our research reveals that this tidy cycle looks rather different in practice.
Interviews with 15 international development organizations suggest that learning does take place – but not primarily where the RBM cycle assumes it does. Rather than predominantly occurring at the end of the cycle, after results have been reported and validated, learning often occurs much earlier during the initial stages of planning and forecasting. These findings challenge one of RBM’s core assumptions.
In the early planning phase, IDOs develop key strategic tools such as Theories of Change – frameworks that map how activities are expected to lead to desired outcomes. Our research shows that this process is far more than a technical box-ticking exercise: IDOs draw on a range of information sources, including research evidence, evaluations, and staff experience. These processes are often dynamic and collaborative: staff come together to interpret evidence, share experiences, debate assumptions, and imagine future scenarios.
By contrast, the later stages of the RBM cycle – particularly reporting and data validation – are perceived as less conducive to learning. These processes are frequently experienced as bureaucratic and burdensome. Staff are required to collect and verify data that may be incomplete, difficult to access, or poorly suited to capturing the complexities of development work. The resulting reports tend to follow rigid templates, leaving only limited room for reflection or dialogue. As a result, reporting often becomes an exercise in compliance rather than a driver of learning.
This mismatch has real consequences. Funders require results reports backed by verified performance data. As a result, aid organizations have often created large departments dedicated to reporting and data verification. However, if IDOs invest in reporting systems on the assumption that they drive learning – whereas, in reality, learning happens elsewhere – then resources may be misallocated.
Meanwhile, the processes that actually appear to support learning – particularly the early, collaborative stages of planning and the uptake of different forms of information sources – often receive fewer resources. The same applies to the use of diverse information sources, including engagement with researchers, evaluations, field observations, and local knowledge.
Our findings also highlight the dual role of Theories of Change. Beyond guiding internal thinking, they serve as powerful communication tools. Because they are often shared externally, they help organizations present a compelling narrative about how their work is expected to generate impact. This may, for example, be used to approach potential funders.
What’s next for development research?
Studies so far are often based either on a few organizations or on a few interviews within these organizations. Given that different types of organizations – bilateral, multilateral, and philanthropic or non-governmental – may face distinct constraints that shape how learning occurs, more research is needed. Still, the implications of our research are clear.
If development cooperation is to remain credible, it must show that it learns from successes as well as from failures. But learning cannot be engineered simply by collecting more data or producing more reports. It depends on when, where, and how people engage with information. At present, some organizations may be looking for learning in the wrong place.
The challenge, then, is not to abandon RBM, but to rethink it. That means shifting the focus from retrospective reporting based solely on performance data toward forward-looking reflection based on a wider range of information sources, including research evidence. In doing so, development organizations may become not only better at demonstrating impact, but also at achieving it.
