The Coming Humanitarian Data Drought

A declining data footprint

Aid cuts have left hundreds of thousands without food aid and other vital humanitarian services. Although major impacts have been documented by researchers and the media, the damage to the humanitarian sector’s ability to track and monitor global food crises has received less attention. Each year, bodies like FEWSNET and the Integrated Food Security Phase Classification (IPC) track emerging food crises around the world, helping direct scarce resources to those who need them most. Their analysis is underpinned by data collected by the World Food Programme (WFP), based on interviews with over a million survey respondents across the world. But WFP’s data collection efforts were another victim of last year’s cuts, and without that information, the reduced finance left for humanitarian response will be less effective in saving lives.

Agencies like WFP (where one of us, Jean-Martin, works) have done everything in their power to continue reporting on the major food crises in the world, perhaps giving a sense of business as usual. But the frequency and depth of data collection has suffered in an increasing share of the 60 countries covered by these information systems.

Primary data collection is down, as WFP offices have cut down on costly surveys. Only 800,000 WFP interviews took place in 2025, down from 1.1 million in 2024. Only 18 WFP offices conducted real-time monitoring in 2025, down from 35 in 2022, making it more challenging to target aid in risky, rapidly evolving humanitarian contexts. WFP still monitored prices and functionality of over 2,000 markets in 2025, similar to 2024, collecting information that’s essential for early warning systems. However, the number of times enumerators visited these markets dropped by a third in 2025.

Image

WFP data teams. Photo credit: WFP/Arete/Utaama Mahamud

To cope with smaller budgets, WFP data teams have also reduced the frequency of household data collection rounds from two to one per year. Some geographies thought to be less vulnerable are no longer monitored at all, raising the risk of overlooking local crises. WFP has also reduced sample sizes and aggregated data, resulting in less granular estimates of household food security. The data footprint will shrink further as budgets fall even more in 2026.

WFP’s partners have faced similar constraints in acquiring primary data. The SMART surveys that measure child malnutrition—usually led by UNICEF or nongovernmental organizations in partnership with national ministries of health—are now less frequent than in the past. In Sudan, fewer than half of the SMART surveys planned for 2025 actually took place, despite the country’s state of critical famine.

The importance of data, and the need for funding and improvement

WFP primary data feeds into a host of planning processes that are essential for humanitarian coordination: the IPC, West Africa’s Cadre Harmonisé, the Joint Interagency Assessment Framework, the World Bank’s Joint Monitoring Reports, and the UN’s Humanitarian Needs and Response Plans. That means the shrinking data footprint—for WFP and other agencies—has ominous, system-wide implications. Fewer surveys, smaller samples, and entire geographies going unmonitored mean that the warning signs of hunger may go unseen. Analysts may misread vulnerabilities, and responses could be misaligned with needs on the ground, particularly for high-vulnerability groups that are complex to survey such as displaced people. Donors have dramatically slashed support across the aid system, and many have talked about the need to “hyper prioritize” targeting aid to the neediest populations and most impactful programs—prioritization which is dependent on good data.

The global food crisis early warning system is a public good, and a relatively cheap one. Funding for data collection is just a fraction of a percent of overall humanitarian budgets—for instance, all of WFP’s surveys and other data collection cost about $30 million per year. Donors looking to maintain accountability and targeting precision should treat this infrastructure as a priority, whether through existing bilateral channels, multilateral commitments, or philanthropic partnerships.

The sector can also use new tools to do better with less, in part due to advances in artificial intelligence. For example, thanks to a collaboration with Google Research, WFP is now able to forecast households’ food consumption 30 days in advance with 95 percent accuracy. This will help the organization leverage the more limited data available and promote early responses.

Beyond forecasting food consumption, WFP in collaboration with Singapore’s Nanyang Technological University can now automate natural disaster damage assessments using publicly available radar imagery—at a fraction of the cost of using paid high-resolution satellite imagery. The method helped put detailed damage maps in the hands of aid workers in the immediate aftermath of the 2025 earthquake in Myanmar.

New datasets can also help better track displacement and vulnerability from emergencies, which enables more timely delivery of aid. Last year, during a sudden-onset emergency, WFP worked with Flowminder to identify real-time displacement patterns inferred from mobile network operator (MNO) data. The data helped WFP reach 70,000 newly displaced people with emergency rations after they were located using MNO data. The use of MNO data to target aid has been well studied, and has been adopted in other programs like the Togolese government’s Novissi cash transfer program and the DRC’s COVID-19 social protection program. These government programs were run in partnership with the nonprofit GiveDirectly and each country’s MNOs, demonstrating the feasibility of executing such innovative approaches with stakeholders across government, civil society, and private companies.

To do better with less:

1. Humanitarian agencies should lean into their data collaborations with private sector companies and academic institutions to replicate the above examples at scale. The new Data for Action Alliance can support this effort, serving as a data broker that protects privacy and humanitarian principles.
2. Humanitarian agencies can do more with each other and should commit to sharing crisis data with one another in real time—an important agenda item of UN80, the institution’s modernization plan.
3. Humanitarian agencies should pool expertise in data science and analytics that can be costly and hard to find—another key agenda item of UN80.

Although new methods and novel data allow WFP and its partners to do more with less, the best algorithms can only be as good as the information they're given. Primary data collection—interviews of vulnerable families by WFP and its partners—are the foundation of efforts to spot and predict food crises. While the humanitarian sector innovates, donors should step in to prevent a collapse of primary data collection.

Jean-Martin Bauer is the Director of Food Security and Nutrition Analysis at the World Food Programme.

With thanks to Charles Kenny and Tim Ohlenburg for review and comments.