Intro

GiveWell’s primary focus has always been researching, identifying, and directing donations to programs we believe will do the most good. When GiveWell first started, we approached this by looking for organizations that were already delivering highly cost-effective, evidence-backed programs and directing funding to those programs. Over time, we were able to focus further upstream by first identifying highly cost-effective programs and then supporting the development of organizations to deliver them. 

We’ve been able to take an even more expansive view as our research team doubled in size over the last several years. In addition to our core grantmaking, we’re now funding an increased number of grants designed to provide information that we think will help us direct more funding to highly cost-effective programs in the future. This includes things like generating research about program effectiveness, scoping new promising programs, and piloting program variations.  

GiveWell has long made some grants aimed at improving our knowledge base, but this work has now grown substantially and become more systematic. In 2025, GiveWell made 18 grants, totaling approximately $39 million, that were aimed specifically at getting more information to improve future funding decisions. In our latest podcast episode, GiveWell CEO and co-founder Elie Hassenfeld speaks with Program Director Julie Faller about these “value of information” grants.

Elie and Julie discuss:

• Testing variations on cash transfers to improve cost-effectiveness: Following an initial scoping grant, GiveWell recently funded GiveDirectly to pilot three program variations aimed at increasing the economic impact of its flagship program, which provides unconditional cash transfers to very poor households. Each pilot tests a different approach to increasing impact: one provides grants to local businesses ahead of a cash transfer rollout, another targets transfers to the poorest young adults, and the third pairs transfers with footbridge construction to help remote communities access markets and services. Through these pilots, we expect to learn more about the feasibility and potential cost-effectiveness of the program variations, potentially leading to opportunities to help people in need even more.  
• Testing a delivery model for diarrhea treatment: Oral rehydration solution—an inexpensive, effective treatment for diarrhea-related dehydration—is used by far fewer children than could benefit from it. To explore whether door-to-door delivery could increase uptake, GiveWell funded a large randomized controlled trial with the Clinton Health Access Initiative in Bauchi, Nigeria. Initial results are promising, indicating that 3,000 community distributors were able to reach about 80% of households with young children—and that ORS usage increased as a result. GiveWell is now considering whether to fund this model at scale, potentially with multiple implementers.
• Collecting better data on a nutrition program in India: GiveWell has supported Fortify Health, which works to address widespread anemia in India by partnering with flour millers to fortify wheat flour with iron. GiveWell recently funded a household survey in six Indian cities to better understand how much fortified flour people are actually consuming, and who in the household is eating it. The survey provided more information about several parameters that, taken together, allowed us to resolve some uncertainty about the program’s cost-effectiveness. We used what we learned to inform our decision to renew funding for the program.  

These examples reflect a longer-term shift at GiveWell—from an organization that primarily evaluated existing programs to one that increasingly generates the evidence needed to improve its grantmaking and fund more impactful programs in the future.

Visit our All Grants Fund page to learn more about how you can support this work, and listen or subscribe to our podcast for our latest updates.

This episode was recorded on March 13, 2026 and represents our best understanding at that time.