Picture a budget meeting in a finance ministry. Two programs are on the table, and there is money for one. Both come with strong evaluations. Both reached their target populations. Both improved the outcomes they set out to improve.
The evidence says both programs work. It does not say which one to fund.
That is the question the sector rarely answers: what happens after impact?
Not what happens after the evaluation is filed, but what happens after the change itself: once a child is healthier or a harvest is larger, what does that set in motion, and who needs to know?
We have never had more evidence. Organizations can show, with growing confidence, whether a vaccination campaign reaches children, whether a nutrition program improves outcomes, or whether an agricultural innovation raises yields. But for policymakers, financiers and government leaders, evidence of impact is where the conversation begins. The harder task is understanding how those results connect with the goals they are accountable for: participation in work, productivity, resilience, inclusion and long-term prosperity.
Too often, development evidence stops one step short of that story. We call this the pathway gap: the distance between what an evaluation demonstrates and what a decision-maker ultimately needs to understand.
The Question Evaluations Were Not Built to Answer
Most evaluations are designed to test effectiveness. Did the intervention improve outcomes? Did it reach the intended population? Was it delivered efficiently and cost-effectively? The sector has become increasingly sophisticated at answering these questions.
Decision-makers face a different set of choices: where limited money goes, what gets scaled, and how competing priorities are balanced. Effectiveness evidence informs those choices, but rarely settles them.
Research on evidence use points in the same direction. A systematic review of 145 studies across 59 countries found that timely access to relevant evidence and relationships with policymakers were important to evidence use, with timing and opportunity among the most frequently identified barriers (Oliver et al., 2014). A survey of more than 1,600 civil servants in Pakistan and India similarly highlighted that simply making evidence available does not necessarily mean it will influence a decision (EPoD, Harvard Kennedy School).
The lesson is not that decision-makers need more findings. It is that findings need to arrive in a form that helps answer the decision in front of them.
Impact Is the First Link, Not the Whole Story
Health shapes who can work, whether children learn and how households absorb shocks. Nutrition can influence cognitive development and later productive potential. Digital access can remove barriers to participation. Agricultural innovations can affect not only yields but livelihoods, income stability and market participation.
But these wider effects should not simply be assumed. The more useful question is: through what channel could the initial impact travel? A health intervention, for example, may matter through fewer days lost to illness, lower healthcare expenditure, improved school participation, better learning, greater labour-force participation or changes in household behaviour. Each pathway involves a different mechanism, requires different evidence and may matter to a different decision-maker.
School-based deworming in Kenya illustrates why the endpoint of an evaluation matters. The original research found improvements in health and school participation, including benefits extending to untreated children in treatment and nearby schools, while short-term academic test scores did not show similarly clear gains (Miguel and Kremer, 2004). Years later, researchers returned to the same experimental population and examined outcomes that became relevant only as participants moved into adulthood, including education, work, earnings and consumption (Baird et al., 2016; Hamory et al., 2021). The lesson is not that every early intervention necessarily produces long-run economic gains. It is that what an intervention means may become clearer when outcomes are revisited over a longer horizon.
The same principle applies elsewhere. Childhood nutrition may influence health first and learning later. Better learning may eventually affect skills and participation in work. Improved access to financial or digital services may initially change service use but later influence economic participation or household resilience. The challenge is to understand where those pathways are supported by evidence, where they remain hypotheses and where they stop.
The Visibility Problem
Broader effects may extend well beyond the immediate outcome. But they are not automatic, and they are often neither traced nor made visible.
When evidence ends at the first outcome, an intervention may enter a funding conversation with only part of its significance visible. Its contribution to household welfare, productivity, public expenditure or longer-term development objectives may remain outside the analysis - not because those effects necessarily exist, but because the question was never investigated.
The Cost of Hunger in Africa studies offer one example of what changes when a wider pathway is examined. They estimate the economic consequences of child undernutrition through channels including additional health expenditure, lost schooling and lower productivity. Importantly, the analysis reframed nutrition not only as a health issue but as an issue relevant to planning and economic decision-making.
The same body of evidence can also look different when viewed through a fiscal lens. A later assessment of the Kenya deworming evidence compared direct and education-related public costs with estimated future tax revenue associated with higher earnings, concluding that the fiscal gains could offset those costs under its assumptions (Policy Impacts). The point is less the specific calculation than the shift in perspective: evidence initially generated around child health and schooling can become relevant to a finance ministry for a different reason.
That is the pathway gap in one line: evidence describes outcomes; decisions require an understanding of pathways.
Leaders need to know not only what changed, but what that change may set in motion, for whom, over what period and under what conditions.
The Pathway Test: Six Questions After "Did It Work?"
“Did the intervention work?” remains essential. But it can be followed by six questions that help establish whether the evidence is sufficiently connected to a decision:
- Constraint. What challenge or barrier was the intervention trying to remove?
- Change. What actually changed for people, households or systems?
- Channel. Through what mechanism could that change produce wider social, economic or institutional effects?
- Distribution. Who benefited, who may not have benefited, and how were those effects distributed?
- Conditions. What needs to hold for the benefits to persist, expand or travel further?
- Contribution. How might these effects contribute to broader development objectives, and how confident can we be in that connection?
The first questions establish what changed. The middle questions investigate how and for whom the effect may travel. The final question keeps the interpretation disciplined.
Not every pathway will prove meaningful. Implementation capacity, financing, market conditions, geography, institutions and time can strengthen, weaken or interrupt the route from an intervention to a broader outcome. The purpose is therefore not to manufacture a larger impact number. It is to understand how far the evidence genuinely travels.
Four Lenses for Taking Evidence Beyond the Evaluation
There is no single analytical sequence that every intervention needs to follow. The appropriate analysis depends on the decision being made, the evidence available and how far the effects can credibly be traced. Four lenses can help structure that enquiry.
1. Evidence: What Do We Actually Know?
Start by separating established findings from assumptions. What outcome was observed? How strong is the causal evidence? Does it apply to the population and setting relevant to the current decision? Are important effects measured directly, inferred from other studies or simply expected? A strong pathway begins with clarity about what the existing evidence can - and cannot - support.
2. Mechanism: How Could the Effect Travel?
Identify the channels connecting the immediate result with wider outcomes. For health, these might include healthcare expenditure, school participation, learning, productive time or household behaviour. For agriculture, they might involve yields, income, prices, labour demand and market access. For digital services, they might include information costs, time savings, service access or participation in formal systems. The objective is not to trace every conceivable effect, but to identify the few mechanisms that are material to the decision.
3. Context: What Could Amplify, Limit or Redistribute the Effect?
An intervention does not operate in isolation. Delivery capacity, complementary infrastructure, labour-market conditions, household behaviour, financing arrangements and institutional constraints may all influence whether an initial gain becomes a broader one. Distribution matters too: an average improvement may conceal very different experiences across income groups, locations or populations.
4. Decision: What Does the Evidence Change?
The final test is not whether the analysis produces another indicator. It is whether it helps someone make a better choice. Does the evidence alter the case for scaling an intervention? Reveal an implementation constraint? Change which population should be prioritised? Show a tension between long-run value and near-term affordability? Or identify an evidence gap that should be resolved before a larger commitment is made? The endpoint is not necessarily a monetary valuation or a macroeconomic estimate. It is decision relevance.
Helping Evidence Travel Further
Budgets are tight, needs are growing, and leaders are under pressure to back interventions that deliver meaningful and lasting results. In that environment, the next frontier may not simply be producing more data. The sector already generates substantial amounts of evidence. The challenge is helping the evidence we have travel from the report to the decision.
A case from our field work shows what that can look like. In one sub-national health system in West Africa, maternal deaths were rising in one area. Leadership suspected a stock-out of a critical medicine, but commodity data showed full supply. Every dashboard suggested the situation was normal. The explanation emerged only when different information already held by the system was considered together: whether the cold chain worked, whether facilities had power and whether staff were trained.
Nothing about the problem reflected a shortage of data. It was a shortage of data brought together in a form that could answer the decision at hand. Analytics produce a signal. People convert it into a decision. That conversion is where evidence earns its value.
What Closing the Pathway Gap Looks Like in Practice
Across our work with foundations, governments and their partners in Africa and South Asia, several practices repeatedly matter:
- Start with a clean results chain. Outputs should not be presented as outcomes, and indicators need a clear relationship to the change an intervention is trying to achieve.
- Design measurement around the decision, not only the report. Start with who will use the evidence and what choice it is expected to inform. That determines which outcomes need to be followed and for how long.
- Separate observation from interpretation. Be clear about what was directly measured, what is supported by external research and what remains a hypothesis about how effects may travel.
- Claim contribution carefully. Wider development outcomes are shaped by many forces. Evidence should show where an intervention plausibly contributes without implying that it independently caused every downstream change.
- Make uncertainty visible. Longer pathways inevitably contain more assumptions. Those assumptions should be explicit rather than hidden behind a single headline estimate.
- Keep the analysis usable. Sophisticated analysis creates little value if the decision-maker cannot understand what it means. A small number of well-evidenced pathways may be more useful than a highly complex model that obscures the choice in front of the user.
Evidence remains one of the most powerful tools the development community has. Its value does not end when an evaluation closes or a metric is reported. It lies in helping decision-makers understand what changed, through what channel that change may travel, where the pathway is supported by evidence, what could interrupt it, and what that means for the decision in front of them.
Proving that an intervention works is important.
Understanding what happens after impact may be even more important.
