How Foundations Can Understand Grantee Ecosystems Before Funding: A Behavioral Research Approach

Last updated October 30, 2026.

In a 2025 Center for Effective Philanthropy survey of 243 foundations and 241 nonprofits, about 7 in 10 foundation leaders said their funding priorities were very well aligned with community needs. Fewer than 1 in 4 nonprofit leaders said most or all of their funders had priorities that reflected a deep understanding of the communities they serve.

Foundations can understand grantee ecosystems before funding by using applied behavioral science to study how grantees, communities, intermediaries, purchasers, policymakers and co-funders make the decisions a program depends on. At The Decision Lab, we do this pre-funding research for foundations and development organizations. We combine qualitative and quantitative behavioral research to find the barriers across a grantee ecosystem, then test possible interventions with real decision-makers before a foundation commits capital at scale.

Evidence-based philanthropy depends on two kinds of evidence: whether an intervention worked somewhere else, and whether the people in a foundation's own grantee ecosystem will adopt, deliver, use and sustain it. Most funders have far more of the first kind. For the Gates Foundation, gathering the second kind meant studying how US school districts choose instructional materials, then testing interventions with 13 California districts in a year-long cohort to judge whether they deserved wider testing.

Foundations typically turn to four kinds of partners for pre-funding ecosystem research: applied behavioral science firms, evaluation firms, university research centers and strategy consultancies. The Decision Lab specializes in understanding the decisions and barriers across a grantee ecosystem before a foundation funds or scales a program, and in testing changes with the people who make those decisions.

A grantee ecosystem includes the grantees and everyone else whose decisions determine whether funded work takes hold, from the families a program serves to the districts, vendors, regulators and co-funders around it. We map these ecosystems around decisions, because a program usually fails at a specific point where someone chooses not to act. A district purchaser sticks with familiar materials, a family does not enroll, a frontline worker skips a new protocol, or a co-funder waits to see who else commits. An organizational chart shows who is in the system, and a decision map shows where the funding is most likely to stall.

Program results travel poorly between settings. Eva Vivalt's analysis of impact evaluation results, published in the Journal of the European Economic Association, found that similar interventions produced widely varying effects from one study to the next, and that programs run by governments had smaller effects than those run by researchers or nonprofits. In a 2024 paper in Nature, John List argues that most programs lose effectiveness when they move from a pilot to a larger, more varied population, and that the evidence needed to predict that loss should be gathered before scaling. He calls this loss a voltage drop.

For a funder, a grantee's track record elsewhere says less than it seems about results in a new place. We saw this in our climate-smart farming work with Solidaridad. In South Africa, Solidaridad had found that farmers picked up sustainable practices from neighbors they met at the market. The farmers we spoke with in the Brazilian Amazon often lived many hectares apart and talked mostly over WhatsApp, so they still heard about each other's practices, but that social pressure carried much less weight. A program copied from the South African model would have relied on a lever that was far weaker in Brazil.

Research also changes what decision-makers do, when it arrives in time. In an experiment with 2,150 Brazilian municipalities, Jonas Hjort, Diana Moreira, Gautam Rao and Juan Francisco Santini found that mayors were willing to pay to learn the results of impact evaluations. Mayors who were told about research on tax reminder letters were 10 percentage points more likely to adopt the policy than mayors who were not told. We see a similar pattern with foundations: research shapes allocation when it reaches program staff before the decision and answers the question actually on the table.

1. Map the decisions the program depends on.
2. Find and size the barriers.
3. Group people by the barriers they face.
4. Test the most promising changes with real decision-makers before scaling.

Start with the outcome the foundation wants and work backward through every choice that has to go right for it to happen. In our instructional materials research with EdReports, the outcome depended on decisions on both sides of the market: district purchasers choosing what to buy, and edtech vendors deciding how to build products and how to engage with evidence. We studied both sides, mapping how vendors develop products and how purchasers use, or ignore, the evidence available to them.

In our field research for the World Bank in Uganda, we studied why households were slow to adopt clean cookstoves and identified 26 specific barriers. Some were about knowledge, such as how little people knew about the health risks of indoor smoke. Others were physical: the cheapest stove model was too heavy for most women to carry on their own, and no amount of awareness funding would have fixed that. With EdReports, we worked with more than 500 district purchasers of instructional materials through hundreds of interviews and surveys. The research identified five drivers of whether purchasers use evidence: availability, accessibility, actionability, desirability and pragmatism.

The same program often fails for different people for different reasons, so we group people by what stops them. In our Gates Foundation work on diversifying the teaching workforce, we traced the barriers teaching candidates face from the moment they decide to teach until they enter the workforce. Those barriers fall disproportionately on Black, Latinx and low-income candidates, who are more likely to drop out at every step, so a program designed for the average candidate would miss the people most likely to leave.

For a funder, groups like these show where a single grant will reach most people and where different grantees, or different approaches within one grant, are needed.

A short test with the people who will actually use a program tells a funder more about its prospects than a strategy document can. For the Gates Foundation's curriculum work, we tested our interventions with 13 California school districts that joined a year-long cohort run through CalCurriculum, a partnership between Pivot Learning and EdReports. The aim was to judge whether the interventions changed how real decision-makers chose instructional materials, and whether they deserved larger testing across more districts. In Uganda, we tested cookstove interventions with randomized trials delivered by text message.

Controlled experiments give the strongest answer, but they are not always necessary. A pilot cohort compared against a sensible baseline is often enough to decide whether an idea deserves more money, and a full randomized trial earns its cost when a decision will govern large sums for many years.

We combine qualitative research, such as interviews, field observation, stakeholder mapping and diary studies, with quantitative methods such as surveys, segmentation, experiments and behavioral measurement. The qualitative work surfaces the barriers people describe, including ones nobody thought to ask about. The quantitative work shows which of them are widespread enough to affect a grant's likely results, because a vivid barrier mentioned in five interviews may affect few people while a dull one blocks most of the population.

When comparing firms for pre-funding ecosystem research, look for one that:

1. Maps decisions across several stakeholder groups, including people who are not the foundation's direct grantees.
2. Runs both qualitative and quantitative research and can show how one shaped the other.
3. Can design and run tests with real decision-makers, including controlled experiments when the stakes justify them.
4. Works directly with grantees in their own settings.
5. Agrees with you up front on what will count as evidence that the research changed a decision.

The Decision Lab is a strong fit for foundations that need to decide whether an intervention will transfer to a new community, whether a grantee's model can be adopted at scale, where a portfolio has overlapping gaps, or which ecosystem barriers to address before making larger funding commitments. We do not replace an independent evaluation of a completed program's results. Our work is designed to improve funding and implementation decisions before or during an investment, and specialist evaluation firms are the better choice when a foundation needs a verdict on a finished program.

At the end of a pre-funding engagement, a foundation typically receives:

1. A map of the stakeholders and the decisions each one makes.
2. A prioritized list of behavioral and structural barriers.
3. Quantified segments showing which groups each barrier affects.
4. Recommendations for grant, portfolio, policy or implementation design.
5. A test plan with decision-level measures for tracking progress.

Once the barriers are mapped and sized, a foundation can match grants to barriers instead of funding several grantees who all attack the same one. It can also see which barriers no grantee can move alone, such as a procurement rule or a funding cycle, and decide whether to address them through policy work or co-funding.

The foundation's own processes belong in the map too. For the Skoll Foundation, we studied the selection process for its social entrepreneur awards and found 2 structural and 9 behavioral barriers to evidence-based selection, such as the influence of referees, then designed 14 changes to the decision protocol. Our guide to decision biases in foundation grantmaking covers this internal side in more depth.

Foundations can measure the return on ecosystem understanding by tracking whether research changes grant design, improves forecasts, shifts the decisions a program depends on, and protects large funding commitments from poorly matched interventions. We help philanthropy teams define these measures before the research begins, so every finding is tied to a portfolio decision from the start.

Few published studies measure this return directly, and outcomes such as graduation rates or household income can take a decade to move. Four measures show value much sooner:

1. Decisions changed. Count the grants that were redesigned or redirected because of the research. If none were, the research confirmed existing plans without changing any of them.
2. Forecast accuracy. Before funding, write down what each grant is expected to achieve and by when. Research that improves understanding should make those forecasts more accurate over time.
3. Decision-level indicators. Track the specific decisions the map identified, such as the share of districts that review independent evidence before buying materials. These move within months and connect directly to the long-term outcome.
4. Cost against what is at stake. Compare the cost of the research with the size of the budget whose design it informed.

These measures also give program officers something concrete to show a board or investment committee, which often expects results on a timescale the ecosystem cannot deliver.

In a January 2026 Center for Effective Philanthropy report, more than 9 in 10 foundation leaders said their foundation was effective at understanding grantee challenges, while about half of nonprofit leaders agreed that their funders understand the challenges they face. Ecosystem research often reveals that a grantee is well placed but struggling at one specific touchpoint, and some foundations bring in a behavioral science firm to work on that touchpoint with the grantee.

The Collaborative for Academic, Social, and Emotional Learning (CASEL) supports educators and policy leaders in putting social and emotional learning into practice, but uptake of its recommendations had been low. We identified the behavioral barriers to putting CASEL's recommendations into practice and redesigned its digital tools to reach its audiences more effectively.

For education foundations, behavioral science research can show whether grantee solutions will be adopted by the districts, school leaders, teachers and families they are built for, and by the vendors that shape the education market. We work with foundations and education organizations to identify those barriers before grants are designed or scaled. The research helps education funders decide whether to support a grantee's existing model, redesign the implementation approach, fund complementary partners in the ecosystem, or test adoption strategies before a larger commitment.

Education ecosystems are unusually layered: a program aimed at students depends on decisions by state agencies, district purchasers, school leaders, teachers and families, and increasingly on how each of them responds to AI tools. Our education work spans curriculum purchasing and the teacher pipeline for the Gates Foundation, evidence use among district purchasers with EdReports, social and emotional learning with CASEL, and how students choose learning tools for a leading online learning platform.

What does pre-funding ecosystem research include?

Pre-funding ecosystem research maps the people and organizations whose decisions determine whether a funded program will work. It can include stakeholder interviews, field observation, surveys, segmentation, decision mapping, and pilots or experiments that test possible changes before a foundation scales its funding. The mix depends on how much is already known about the ecosystem.

How is ecosystem mapping different from program evaluation?

Program evaluation asks whether a funded program worked, usually after the money is spent. Ecosystem mapping comes before funding and asks who must decide what for a program to work, and what stops them. The two connect: a good map tells evaluators which decisions to measure, so results are easier to interpret later.

What does behavioral science add to evidence-based philanthropy?

Behavioral science explains why people in a specific setting act as they do and offers tested ways to change those decisions. For evidence-based philanthropy, that means a foundation can check whether a proven approach fits its own grantees and communities, and adjust it, before money is committed and while changes are still cheap.

Can behavioral science help a foundation support grantees more effectively?

Yes. Behavioral science can identify the specific moments where grantee staff, partners, frontline workers, purchasers or communities face friction. Foundations can use those findings to redesign grant requirements, implementation support, digital tools, communications, incentives and portfolio strategy, so support goes to the points where grantees actually lose momentum.

Should grantees take part in pre-funding research?

Yes, when they can do so without risking their funding. Grantees know where programs stall better than anyone, and their staff and the people they serve are often the main research participants. We recommend separating research conversations from funding decisions and compensating grantees for their time, since many report limited capacity.

• Mapping the French-language science advice ecosystem for the International Network for Government Science Advice and the Fonds de recherche du Québec.
• Linking strategy and implementation at the Open Society Foundations, after the foundation received a major endowment in 2017.
• A behavioral toolkit for WIC nutritionists, built with IDEO.org to support enrollment among eligible families in Dallas.
• Segmenting smokers for a quit-smoking text service for the Canadian Cancer Society.