It is a justifiable commonplace that data on outcomes is key to effective aid: for targeting, delivery, monitoring and evaluation. Last year certainly provided additional evidence of the need for data to target assistance: because of cuts to data collection we simply didn’t know where a lot of food insecure people were in 2025. Without HIV testing, you can’t find the people who need lifesaving antiretroviral drugs. And compared to audit efforts following paper trails of receipts, outcome data is a far more useful approach to uncover waste, fraud, and corruption. This same data is also vital for monitoring and evaluation: without a measure of something working, we can’t know what has worked. But on top of all of that, last year demonstrated again that we need quality outcome data as a powerful tool to justify and sustain aid.
The power of data for advocacy was made clear by the Trump administration's efforts to suppress it last year. In the case of the US global HIV program PEPFAR, there should have been quarterly reports in 2025 on what was happening to service delivery, but the State Department took reporting offline (surveys run by academics filled some of that gap and the system is now partially back online). Again, as famine loomed in Somalia, Afghanistan, and Yemen, all subject to particularly deep cuts in US support, the State Department stopped famine early warning coverage of those countries through the Famine Early Warning System Network (FEWSNET) (thankfully, the decision was reversed after complaints).
And compare the treatment of PEPFAR over the last year to much of the rest of the foreign assistance portfolio. PEPFAR is credited with saving more than 26 million lives over the past two decades, an estimate that has been quoted approvingly by the Trump administration. That’s a modeled number, but it is one with a comparatively high degree of confidence behind it precisely because PEPFAR has long run a considerable data collection exercise, so that there can be simply no doubt that the program was delivering lifesaving assistance. It was also comparatively straightforward to forecast where we would see the greatest impact of any significant pause in funding even without access to the data the administration took offline. This strong foundation of data helped considerable and effective advocacy on behalf of PEPFAR on and off Capitol Hill.
Of course, PEPFAR isn’t nearly where it was: thanks once again to data on delivery of services, we can say that the number of people on PEPFAR-supported antiretroviral treatment fell by nearly two million between FY2024 and FY2025, overwhelmingly concentrated in South Africa. Coverage for harder-to-reach groups, prevention, and testing all took significant hits. And the future of the PEPFAR program under the administration’s America First Global Health Strategy is anything but assured.
Nonetheless, it is in a considerably better place than other parts of the US foreign assistance program, many of which are in sectors where it is simply difficult to generate similarly compelling outcome metrics: some of the deepest cuts in USAID programs were in activities related to private sector development, policies and regulations, private investment, and political competition. These were activities that had little support in the administration, but as importantly the program cuts garnered comparatively little opposition from Congress members who continued funding them.
With regard to humanitarian assistance, data covering food security that is vital for targeting is still being collected, if at a reduced level. But data on outcomes has long been patchy, a bar to all three of monitoring, evaluation, and advocacy. The evaluation of humanitarian assistance is fragmented enough that it was difficult to provide (even) a defensible modeled impact of humanitarian finance cuts in 2025. For all that FEWSNET reporting worried the administration, the lack of outcome data may be one factor behind the comparatively steep cuts to humanitarian support compared to health assistance (a 62 percent cut in humanitarian assistance obligations FY24 to FY25 compared to a 34 percent cut for health).
For advocacy in particular, data is not enough. Tim Hirschel-Burns’ list of individuals who have died as a result of US aid cuts relies on the careful reporting of numerous journalists across countries, and the names come with compelling, devastating stories. The statistics cannot equal that, but can suggest how many stories remain untold.
The need is for compelling (seemingly) simple indicators of outcomes or widely accepted proxies. Mortality and antiretroviral coverage pass that bar, but for all of its value, food insecurity likely does not. To echo David Miliband, the sector needs more accountability for outcomes: “healthy births not health consultations, children fed not food parcels tallied.”
It is not easy to measure outcomes of some really good projects that prevent bad stuff from happening or that are oriented at the macro environment where the causal chain is long and partial from assistance to outcome. Stories here are even more important and modeling may have a role. We can’t just give up on aid where it is hard to get to compelling impact measures. But where it is possible we should collect and use those impact measures not only to monitor and improve, but also to sustain foreign assistance. Better data collection can’t wait on revived aid flows, because it will be one of the things that helps deliver that outcome.
