Our interconnected humanitarian data and analysis ecosystem: resilience, reckoning?

In July 2012, then UN Secretary-General Ban Ki-moon called for a ‘data revolution’ and a new era of data-driven, evidence-based humanitarian decision-making, including enhanced coordination and data sharing, and improvements in data collection and analysis. Against a backdrop of the demise of the United States Agency for International Development (USAID) and funding reductions amongst major donors, how far have we come since Ban Ki-moon’s announcement and where do we go next?

Since the 2012 announcement, there has been a multiplication of data-collection approaches and provision of humanitarian analysis and insights in various formats (e.g. reports, dashboards, indices) and a growth in organisations supporting data collection, management and analysis. Looking at two data and analysis repositories alone, Reliefweb hosts over 1 million updates and there are just over 18,000 datasets on the Humanitarian Data Exchange, hosted by the UN Office for the Coordination of Humanitarian Affairs (OCHA) These figures are correct as of 7 April 2025. . These products are used for myriad reasons, from quantifying humanitarian needs and severity, identifying gaps, prioritising response decisions and supporting donor funding allocations. Technological advances and the growing use of ‘big data’ has seen forecast models and AI-assisted analysis tools emerge. Concurrently, there has been enhanced use of weather and climate, conflict, and economic data to support contextual analysis and early warning.

Alongside the growth of humanitarian data, there have been improvements in data coordination and sharing. Whilst not perfect, this has seen enhanced data, including to and from national governments, development of multi-stakeholder analysis approaches and creation of common operational datasets (these datasets currently have a warning due to US stop work orders impacting their workflows). Much of this collaboration goes unnoticed. This article aims to highlight some critical connections within our humanitarian data and analysis ecosystem, as seen in this interactive map. (Note that this map is ever expanding as new connections are identified. Read the information panel on the left-hand side of the webpage to navigate and query the map.)

What are the implications when a data and analysis source comes to a halt? Will we miss the next famine if there are interruptions in data and analysis sources? Will we have to revert to politicised analysis provided by actors to conflict? What opportunities are there to advocate for ‘doing data and analysis differently’ as the wider humanitarian system faces change?

A critical need for multidimensional data and analysis: preventing famine

Famines are complex. Preventing famines, and the most severe acute food insecurity outcomes, requires identifying and tracking climate, conflict, agriculture, economic, socio-demographic and response factors, applying expert judgement and working with probabilistic forecasts in data-poor and access-constrained environments. Access to historical quantitative and qualitative datasets enables the identification of patterns and trajectories into famine. These patterns and trajectories vary across contexts, with no ‘one size fits all’ approach to analysing how factors interact over different timeframes. Ultimately, identifying famine, and acute food insecurity risks, is a data- and analysis-intensive process that has undergone significant development since the 1970s.

A highly visible ‘data and analysis casualty’ of the USAID cuts in 2025 is the Famine Early Warning Systems Network (FEWS NET). Established in 1985, FEWS NET has a global network of analysts and partnerships that supports the identification of emerging acute food security risks. It uses a combination of data sources and scenario-based analysis to ascertain the likely acute food security pathways in a given country. Reviews (2019, 2021) of FEWS NET’s ability to forecast future food security outcomes indicated there were few severe outcomes, including small-scale issues, that were not foreseen, even if challenges persisted over the projection of severe outcomes. Even with indications that FEWS NET will resume in some capacity from mid-2025, monitoring and analysis interruptions risk us not being able to identify emerging severe food security outcomes over the coming months.

Given the multi-dimensional nature of famine risk, it is essential to triangulate sources and consider a range of plausible futures. Linkages between the Integrated Food Security Phase Classification (IPC/CH) and FEWS NET are well established This link was active at the time of writing, but is no longer working due to the current closure of FEWS NET (April 2025). , but nuances in their analysis approaches and raisons d’être can emphasise different factors, seeing differing results. The IPC/CH process heavily utilises UN-led data-collection processes from the food security, health and nutrition sectors. FEWS NET, which also collates this information, utilised more Earth Observation data than the IPC/CH process and was not national government-, but USAID-led. Other analysis, such as the UN Food and Agriculture Organization’s State of food security and nutrition in the world reports, works more closely with developmental indicators, providing a helpful overview of the vulnerabilities that can contribute to adverse food security outcomes in the event of shocks. Having these differing approaches can help provide a more complete picture of differing outcomes and support the identification of areas of intervention aligning to different organisational mandates and expertise. With FEWS NET currently offline, and with the IPC/CH facing funding challenges, will we be left without a way to identify emerging famine risks?

Climate and weather data for early warning, preparedness and anticipatory action

Climate and weather data is found across the sector. To understand where to prioritise the most at-risk communities, you need hazard and exposure data and historical event and impact information (the last of which is provided by the Emergency Events Database, which is partially funded by USAID). To monitor and act ahead of natural hazards, you need weather and climate observations and forecasts. In particular, the early warning, preparedness and anticipatory action fields require access to significant hazard, forecast and observation data, much of which comes from international sources. For example, the US National Oceanic and Atmospheric Administration’s (NOAA)’s National Hurricane Center and Central Pacific Hurricane Center are the official forecasters for hurricanes in the North Atlantic, Caribbean and Eastern Pacific basins. Their seasonal hurricane forecasts are used by organisations to conduct preparedness and early-warning activities, ensuring relief items can be rapidly mobilised and communities and humanitarian personnel are aware of enhanced risk. Short-term forecasts for active hurricanes are used to generate alerts, such as those produced by the Global Disaster Alert and Coordination System (GDACS) and the World Food Programme’s Advanced Disaster Analysis and Mapping (WFP ADAM), which provide estimates on population and critical humanitarian infrastructure exposure. NOAA has already undergone staff cuts under the current US administration, with the fallout impacts on early-warning systems yet to be fully realised.

Climate data is highly interconnected. Seasonal climate forecasts use a consensus approach, where multiple forecasts from different meteorological agencies are combined into a single, agreed-upon forecast ahead of rainy seasons in climatologically similar regions around the world, cascading to national and sub-national levels. These seasonal forecasts require monitoring of slow-moving aspects of our climate, such as fluctuating ocean sea surface temperatures. One example of this is the El Niño-Southern Oscillation (ENSO), which can influence global weather patterns. NOAA’s Climate Prediction Centre (CPC) emerged as an authoritative voice in monitoring and forecasting ENSO fluctuations in the 1990s, something that continues today, with ENSO forecasts produced by the International Research Institute for Climate and Society (IRI) in partnership with NOAA utilised by the Inter-Agency Standing Committee in its ENSO protocols.

Climate collaboration extends across academic, governmental and humanitarian organisations. The Climate Hazards Centre at the University of California, Santa Barbara was established in 2003 in collaboration with the US Geological Survey to support FEWS NET. The humanitarian sector frequently uses the centre’s rain gauge and satellite observation dataset, known as CHIRPS. This dataset underpins many hazard early warning, agriculture condition and humanitarian risk platforms, including globally, regionally (e.g. the Horn of Africa) and at country level (e.g. Burkina Faso), supporting identification of risks to crop development and potentially severe acute food security outcomes. Will US cuts affect the accuracy of seasonal forecasts? This remains unknown.

Quantifying human needs and donor prioritising: people in need

We use a lot of numbers. The quest for ‘evidence-based decision-making’ has often translated into use of data and analysis in donor prioritisation and decision-making frameworks, alongside geopolitical factors. There is frequent use of quantitative data: numbers of people facing ‘Crisis’ (IPC Phase 3) or above acute food insecurity outcomes, the number of people internally displaced, poorly funded humanitarian response plans or contexts considered ‘Very High’ severity. Use of these sources often goes beyond humanitarian teams, crossing into conflict, economic and development decision-making processes.

One number that is often used is the People in Need (PiN) metric, contained within UN coordinated appeals and summarised in the Global Humanitarian Overview (GHO), launched by OCHA annually. We get to this metric through the Joint and Intersectoral Analysis Framework (JIAF). The JIAF was born of donor and humanitarian agency commitments at the World Humanitarian Summit in 2016, to both improve the quality of humanitarian needs analysis and move towards more coordinated assessments. It is a process designed to enable analysis of the varying dimensions of a crisis (context, drivers of need, impacts of those drivers and humanitarian conditions), bringing together sectoral assessments that consider how they relate to each other. After a 2021 independent review Yale University Independent Review Team (2021). Final report: Independent Review of the Joint Intersectoral Analysis Framework (JIAF). , the methodology was updated to enhance cross-sectoral comparison of severity, amongst other changes, with JIAF 2.0 rolled out in 2023.

Production of the PiN and the response plans is a highly data- and analysis-intensive process. In calculating a ‘Joint overall PiN’, the JIAF approach utilises a swathe of indicators from the humanitarian sector, alongside qualitative inputs to derive the contextual components of humanitarian response plans. The PiN is often formulated using data from existing assessments (e.g. IPC/CH, the Consolidated Approach for Reporting Indicators of Food Insecurity (CARI) and health assessments), but also from primary data collection as part of Multi-sectoral Needs Assessments (MSNAs) conducted by REACH. MSNAs have historically been funded by major donors who are now cutting official development assistance, including the US and UK. Whilst debates around the intensive nature of the methodology and reliability of the PiN metric persist, the PiN, both aggregated to global level and for individual countries, remains a highly visible and widely quoted humanitarian metric, both within the sector and news media. With funding cuts, will we see a global PiN in 2026? Does it matter if we don’t?

Resilience, reckoning?

With the challenges facing data and analysis providers across the system, we need a collective assessment of our ‘data and analysis resilience’, alongside wider understanding of how we are using data and analysis in our day-to-day work. Many people, including decision-makers at all levels, may not realise how connected our data and analysis ecosystem is and how many sources there are. Whilst we have seen the creation of, in effect, a pooled fund for data in the Complex Risk Analytics Fund (CRAF’d), it is not enough to sustain funding to a few data and analysis providers if it is decided we truly need everything – there is no such thing as an unconnected source. However, key questions remain. Who are the decision-makers, and what decisions are being made? Do we need everything in the current system or is it information overload for little decision-making gain? Shouldn’t we shift our view of who decision-makers are? The gaps in the ‘types of numbers and analysis used’ between donors, local NGOs and communities are vast – do local actors, who are so key to humanitarian response, find the PiN metric useful, for example?

So, what now? We can work to ensure ‘data and analysis resilience’, protecting our existing ecosystem. Equally, with the turmoil facing the humanitarian sector and calls for a radical redesign of the humanitarian system in light of the demise of USAID and upcoming cuts, this is an opportunity for assessing what we truly need and how we approach some of our data collection and analysis – an ‘analysis reckoning’. If our current approach is ‘impartial’ and ‘needs based’, who has defined what is ‘impartial’ and what ‘in need’ means (recognising that governments have been successful in supressing data and analysis identifying famine and internal displacement)? Are needs and severity of needs really universal? Can we define our questions before collecting data (‘question first’, not ‘data first’)? Can we really boil down human complexities to a single number in the Global Humanitarian Overview? As eyes seem to be turning to ‘big tech’ to provide solutions to funding gaps and data and analysis needs Noting here the inclusion of an Amazon Web Services representative in the panel for the keynote talk on ‘The Future of the Humanitarian System’ at Humanitarian Networks and Partnerships Week (HNPW) in Geneva, March 2025. , can these actors be involved in line with humanitarian principles? Would turning to ‘big tech’ risk replicating power dynamics we already observe with Global North donors?

Quantifying needs and severity is a western construct. As Glasman says, ‘The production and storage of knowledge about populations (archives, files, cadasters, registers, census, identity documents, maps, statistics, etc.) is a key feature of modern European states.’ The current system is rooted in a paternalistic, quantitative donor-driven model, where numbers are king, needs are pre-defined The point on needs being pre-defined was well illustrated by an audience question at an HNPW 2025 session, ‘Does any survey on needs include freedom, liberation or smashing the patriarchy?’, from the 58-minute mark. , and nuanced, expert judgement with use of qualitative data often absent. Do people truly understand what numbers mean, when they are provided without context? Examples here could include: Almost 5 million people in Haiti are experiencing Crisis or above acute food security outcomes (IPC 3+) – out of context this number seems ‘small’ compared to other contexts but as a proportion is almost 50% of the population. Counting trucks as a proxy for humanitarian access is also meaningless – the number of trucks crossing a checkpoint tells you nothing about the last mile, and how aid within those trucks is reaching people. Communities often remain disconnected from the numbers, although a limited number of UN coordinated appeals do now contain testimonials from community members. If the entire humanitarian system becomes more localised, what would our ideal data and analysis ecosystem look like? Who defines data, and the data-collection questions?

Within the early warning and anticipatory action space, is there an opportunity to better advocate for ‘no regrets’ programming that supports decision-making when working with qualitative and probabilistic data? We know decision-makers, in this instance defined as donors, struggle to deal with uncertainty, particularly when working with probabilistic forecasts. If there is a collective move towards managing uncertainty and increasing community inputs in a more positive way, does this lessen the risks? With disruptions to collection and provision of quantitative data, this may have to be the approach taken, alongside use of qualitative, highly contextualised community-led data sources that can rapidly provide information on emerging crises.

Ultimately, the USAID disruption highlights how reliant we are on data and analysis to identify emerging risks and severe humanitarian outcomes, and prioritise programming, alongside how interconnected the data and analysis ecosystem is. Data and analysis are essential to preventing unnecessary suffering, with many data and analysis providers and approaches at risk following the demise of USAID and future funding cuts. Whilst there are now opportunities to consider what communities really need and how data and analysis evolve as part of a ‘data and analysis reckoning’, without ensuring ‘data and analysis resilience’, those who can act might end up making decisions in the dark, or worse yet, not making them at all.

Beth Simons is a Humanitarian Analyst, and has worked for INGOs, the UN and at a donor in various roles, including as a senior humanitarian analyst utilising many of the data and analysis sources mentioned in this article for humanitarian early warning and prioritisation.

This article draws on an ongoing piece of independent research involving systematic mapping of the humanitarian data and analysis ecosystem and investigating the challenges and opportunities presented by ‘indicator culture’, data literacy and the use of simplified metrics in decision-making.