GiveWell’s research doesn’t end once we’ve made a grant. We evaluate a subset of completed grants, comparing what we thought would happen to what actually took place, then try to use what we learn to improve our future funding decisions. Over the past year, we’ve formalized and expanded this work, publishing comprehensive “lookbacks” for select grants.

A recent lookback on grants GiveWell made to fund insecticide-treated net distributions supported by the Against Malaria Foundation (AMF) in the Democratic Republic of Congo (DRC) illustrates the growing capacity of GiveWell’s research team. We drew on multiple independent data sources, funded qualitative interviews to gather more information, and conducted a novel empirical analysis to deepen our confidence.

In this episode, based on a conversation originally aired on GiveWell’s internal podcast for staff,* GiveWell’s co-founder and CEO Elie Hassenfeld provides additional context while GiveWell’s Chief Research and Program Officer Teryn Mattox dives deep into the details with Program Director Alex Cohen and Researcher Steven Brownstone, examining how we conducted the lookback, what we found, and how what we learned may shape our future nets grantmaking.

Elie, Teryn, Alex, and Steven discuss:

• A more expansive and rigorous approach to evaluating past grants. This lookback draws on three independent quantitative sources—AMF’s monitoring data, a recent Demographic and Health Survey (DHS) conducted in the DRC, and an original survey commissioned by GiveWell—alongside qualitative research involving in-depth interviews with people involved in DRC’s net distribution system, from health zone administrators to village focus groups.
• Conducting a novel mortality analysis using DHS microdata. Because net campaigns roll out on staggered schedules across DRC’s provinces, we were able to use the timing of children’s births relative to the date of local net campaigns as a natural experiment. We compared mortality risk for children based on when they were born, and thus the length of time they had protection from a net, and found that the net campaigns reduced the risk of death by around a quarter. That finding provides additional support for the mortality effect estimate we use in our cost-effectiveness models.
• What qualitative research revealed. Interviewers asked people across five provinces in DRC whether households received nets and whether households were using nets—and in cases where they either didn’t receive nets or weren’t using them, why not. Although we heard some anecdotes of misuse or diversion of nets, the data suggested overall that the nets are highly valued by the communities receiving them.
• How durability data could inform campaign design. Our analysis of DHS data confirmed earlier research indicating that nets in DRC degrade before they are replaced through new distributions. As a result, it’s possible that changes in DRC like more frequent campaigns or increased support of routine net distribution through other channels may increase protection.

If you’re interested in learning more about grant lookbacks like this one—and how they’re improving our research and shaping our future funding decisions—we invite you to join our next webinar on June 9. Alex Cohen, who was featured in this episode, and Program Director Julie Faller will walk through our lookback process, what we’re learning, and how we’re applying those lessons to help more people. Learn more and register here.

This episode was recorded on April 22, 2026 and represents our best understanding at that time.

Glossary

Because the conversation in this episode first aired as part of GiveWell’s internal podcast for staff, there are a number of names, acronyms, and other terms that are not explained. To make it easier to follow along, we’ve provided a glossary below.

• all-cause mortality. All-cause mortality measures the total number of deaths from any cause in a specific group of people over a specific period of time.
• AMF. The Against Malaria Foundation, one of GiveWell’s Top Charities, collaborates with national malaria programs and other partner organizations in low- and middle-income countries to distribute insecticide-treated nets.
• CEA. We build cost-effectiveness analyses to assess how much good can be achieved by giving money to a certain program.
• Cox proportional hazards model. The Cox proportional hazards model is used to estimate how much different factors, such as time since an insecticide-treated net campaign, speed up or slow down the time to death. It assigns each factor a “hazard ratio,” which is a multiplier of the baseline risk of dying: a hazard ratio of 2 for smoking means that smokers face double the risk of death at any given moment compared to nonsmokers, all else equal.
• DHS. ​​Demographic and Health Surveys are vast, in-person surveys that ask women to recall their children’s birth and survival histories. This method provides the primary data for mortality estimates in low- and middle-income countries.
• funging. What we call “funging” (from fungibility) refers to the effect of crowding out funding that would have otherwise come from other sources.
• insecticide-treated nets. These nets have been treated with insecticide to deter and kill the mosquitoes that transmit malaria. Distributing insecticide-treated nets, which are then hung over sleeping spaces, can be a cost-effective way of preventing malaria.
• lookbacks. Lookbacks are reviews of past grants published on the GiveWell website that assess how well they’ve met our initial estimates and what we can learn from them.
• Marakuja. Marakuja Kivu Research is a nonprofit organization in DRC that we have contracted with to conduct quantitative and qualitative surveys.
• M&E. GiveWell asks organizations that we fund to share detailed monitoring and evaluation data on their programs to assess the quality of program implementation and whether it is reaching recipients as intended.
• net durability. Insecticide-treated nets decay over time, both through loss of insecticide and physical wear.
• nets team. In internal conversations, this is what we sometimes call our vector control team (see below for definition).
• OnFrontiers. OnFrontiers is a company that sets up interviews with subject matter experts around the world.
• PDM. Post-distribution monitoring data is collected by independent partners funded by AMF. These partners survey a sample of households in the areas targeted by a campaign to assess the presence, usage, and condition of nets over time.
• PLNP. The Programme National de la Lutte contre le Paludisme (or National Malaria Control Program) in DRC plans, coordinates, and implements malaria prevention and treatment strategies.
• PMI. The President’s Malaria Initiative is a US government program to fund malaria prevention and elimination programs.
• regression discontinuity. This is an econometric method that leverages the fact that if outcomes “jump” at a threshold, in this case a geographic border, then the “jump” can be considered the effect of a policy difference across the border.
• STATcompiler indicators. Some indicators derived from the Demographic and Health Surveys (DHS) are available pre-calculated on a website run by the DHS called STATcompiler. Cross-referencing against these pre-calculated values is a good way to validate the analysis of the raw data from DHS surveys.
• vector control team. GiveWell’s vector control team is a research subteam within our malaria research team that focuses on interventions that prevent malaria infections and deaths by targeting the mosquitoes that transmit the disease.