The State of Humanitarian Data

A System Under Stress: Data, Archives, and the Cost of Cutting Memory

Humanitarian data and institutional memory face a compounding crisis in 2025–2026. On one front, the OCHA Centre for Humanitarian Data reports that the share of crisis data that is available and up-to-date has fallen from 74% to 68% across 22 active humanitarian operations — a decline driven overwhelmingly by the largest wave of international funding cuts in recent memory. On another, an entire ecosystem of organisational archives — the documents, field records, WhatsApp threads, and emails that constitute the sector's institutional memory — is deteriorating silently, with no adequate preservation infrastructure for the vast majority of humanitarian organisations.

These two crises are more connected than they appear. The funding cuts that are shrinking real-time data coverage are the same cuts disbanding the enumerator networks, forcing office closures, and eliminating the staff whose knowledge and records will never be formally archived. What is lost is not merely a data point but a body of institutional learning accumulated over years of emergency response. The sector's ability to understand the present is weakening at the same moment its ability to learn from the past is disappearing.

Against this backdrop, two parallel initiatives offer targeted responses. The OCHA Centre for Humanitarian Data continues to manage HDX — a platform used in 237 countries and territories whose nearly 20,000 datasets were downloaded 3.4 million times in 2025. At the University of Manchester, the Humanitarian Archive Emergency (HAE) project and its accompanying Digital Humanitarian Archives Toolkit represent a practical attempt to democratise archiving for resource-poor organisations. Together, they sketch the outlines of what a more resilient humanitarian knowledge infrastructure could look like — if the sector finds the will and resources to build it.

The State of Open Crisis Data: A Fragile 68%

The HDX Data Grids — OCHA's standardised framework for assessing data across 22 active humanitarian operations — provide the clearest snapshot of the sector's data health. As of 31 December 2025, 68% of crisis data is available and up-to-date, down from 74% the previous year. That six-point decline, achieved despite extraordinary mitigation efforts, signals a structural weakening rather than a temporary fluctuation.

The geography of data quality is deeply uneven. Mozambique leads at 90% of sub-categories available and current, followed by Afghanistan, Cameroon, and South Sudan at 85%. At the bottom sit Venezuela (47%), the Syrian Arab Republic (50%), and Burkina Faso (50%) — precisely the contexts where decision-makers most need reliable information. The State of Palestine, where the IPC declared famine conditions in 2025, has only 53% of data available and current.

Two sub-categories registered genuine improvements. Humanitarian needs data rose 30%, driven by the Humanitarian Reset's focus on streamlining planning processes and bringing data online earlier. Food security improved 14% after the Cadre Harmonisé — delayed in 2024 — was published on time in 2025. These positive cases demonstrate that process reforms, not just funding, can shift data quality. They are also important proof of concept as the sector prepares for deeper cuts ahead.

The Human Cost of the Data Drought

Behind every declining percentage point in the HDX Data Grids is a concrete operational decision: a cancelled contract, a closed office, a disbanding enumerator network. The scale of these reductions in 2025–2026 has no recent precedent. UNHCR and IOM have each cut data-related staff by approximately 40%. OCHA has reduced information management capacity in field offices by more than 15%. WFP enumerator visits — the in-person verification that underpins food price and market monitoring — dropped by roughly one-third.

The cascading effects extend well beyond the humanitarian sector. UNHCR's statistical work — which provides socioeconomic data on refugees to governments and development actors — is under pressure. Without adequate resources, government-led solutions that depend on this data may be delayed or misdirected. In Sudan, IOM was forced to cancel contracts for more than 100 enumerators; displacement data collection was disrupted entirely until emergency funding was secured. In Somalia, IOM projects data collection will shrink from nationwide coverage to just 4–5 locations in 2026.

The Bot War: Open Data Platforms Under Siege

While funding cuts constrain the supply of humanitarian data, a new threat has emerged on the demand side: AI bot crawlers deployed by major technology companies to train large language models. In 2025, AI-bot and crawler traffic on the global internet almost quadrupled, climbing from 2.6% of verified bot requests in January to more than 10.1% by the end of Q3. HDX, as a curated open-data platform with structured, high-quality humanitarian datasets, became a prime target.

The platform absorbed periodic server demand surges reaching 20–30 times baseline levels. For the first time, civil society and nonprofits surpassed financial institutions as the most targeted sector for cyberattacks. The very features that make HDX valuable to humanitarian users — open access, no login requirements, rapid download options — also make it maximally attractive to automated scrapers. This creates a structural paradox: openness enables impact but also enables exploitation.

HDX's response — web application firewalls, IP blocking, bot-control rules, geolocation filtering — consumed significant resources. Managing bot identification has become, in the words of the OCHA report, "more of an art than a science." The platform now categorises traffic into three tiers: wanted bots (search crawlers, AI indexers), unwanted bots (scrapers, security probes), and a "grey zone" of unidentifiable traffic. Navigating this complexity will require a strategic reallocation of resources toward infrastructure and specialised skillsets.

The Archive Emergency: What Isn't Preserved Is Lost

Beneath the headline data figures lies a deeper and less visible problem: the systematic failure to preserve the records, field notes, email chains, and operational documents that constitute humanitarian institutional memory. While organisations like UNHCR, ICRC, and IFRC have dedicated archive teams, and others like Oxfam and Save the Children have deposited archives at major universities, the vast majority of humanitarian organisations — particularly smaller national and local NGOs — have no functional archiving practice whatsoever.

Flora Chatt, archivist at the University of Manchester's Humanitarian Archive (founded 2021, currently holding 15 donated collections), frames this as a structural problem compounded by the nature of humanitarian work itself. Fast-paced emergency settings, frequent staff turnover, poor infrastructure for recordkeeping, and the move to digital communications — which require active, ongoing maintenance unlike paper — have created a perfect storm for institutional memory loss.

The shift to digital record-keeping has paradoxically made the problem harder. A box of paper documents, left in a room for decades, remains largely readable. Digital files left for years face a cascade of risks: outdated hardware, obsolete file formats, silent corruption through "bit rot." The technical requirements for digital preservation — format migration, checksum verification, regular monitoring — demand more sustained resource and expertise than physical archiving, at precisely the moment when organisational budgets are being cut.

The Humanitarian Archive Emergency Project and its Toolkit

The Humanitarian Archive Emergency (HAE) project, funded by Elrha's Humanitarian Innovation Fund and running from November 2025 to October 2026, represents the sector's most structured response to the archive crisis. Based at the University of Manchester's Humanitarian and Conflict Response Institute, the project aims to establish a global census of archives at risk, develop reporting mechanisms, create ethical triage frameworks, and shape a medium-term strategy for sustainable rescue. It operates on a one-year timeline — modest for the scale of the problem, but designed to catalyse longer-term action.

The project's most immediate practical output is the Digital Humanitarian Archives Toolkit (beta), an e-learning resource designed to be accessible to any staff member at any humanitarian organisation — regardless of prior archiving knowledge. Its architecture reflects a pragmatic philosophy: the goal is not best practice but good enough practice, scaled to organisations that have almost no dedicated capacity.

Where the Crises Converge: Five Structural Themes

The data availability crisis documented by OCHA and the archive emergency described by the HAE project are not separate phenomena. They are two manifestations of the same underlying structural failure: the humanitarian sector has never treated knowledge infrastructure — data systems, preservation practices, institutional memory — as a core operational asset deserving sustained investment. The following themes recur across all three source documents.

1. Funding Cuts as Existential Threat

All three sources converge on this theme with maximum strength. The OCHA report quantifies it: 68% data availability, 40% staff cuts. The Chatt transcript contextualises it: the DHA Toolkit was "particularly kind of accelerated as well over the past few years" by budget cuts, and "this has become particularly kind of acute very recently in times of severe budget cuts as we are encountering now." The toolkit's design philosophy directly addresses it: built for organisations with "very few resources." The implication is clear — the sector cannot technical-solution its way out of a funding problem.

2. The Asymmetric Recovery Problem

Funding can, in principle, be restored. But much of what is lost in these cuts cannot be quickly rebuilt. Enumerator networks built over years of local relationship-building, once disbanded, require years to reconstruct. Archives deleted or left to decay cannot be recovered. Institutional knowledge held by departing staff dissipates. The OCHA report notes this explicitly regarding staffing cuts; the Chatt transcript makes the same point about archives. The humanitarian sector faces a structurally asymmetric risk profile: cuts are fast and cheap, recovery is slow and expensive.

3. The Equity Gap in Knowledge Infrastructure

Both the data availability data and the archive landscape reveal a sharp divide between large, well-funded organisations and the majority of smaller actors. Large organisations can absorb data staff reductions while maintaining some coverage; small organisations cannot. Large organisations have dedicated archivists and can advocate for their archives' importance; small and local NGOs have neither. The Chatt transcript puts this starkly: "the larger organisations can not only afford to maintain their archives but they can afford to kind of advocate for it." This structural inequality means that the historical record of humanitarian work is disproportionately shaped by large, international actors — a distortion with real implications for accountability and learning.

4. Digital Technology as Both Solution and Problem

The shift to digital work is simultaneously what makes data sharing at scale possible (HDX, Data Grids, the toolkit itself) and what makes preservation dramatically harder. Digital files require active, resource-intensive maintenance. AI bots are both a threat to platform stability and a potentially powerful channel for disseminating humanitarian data. The sector's digital infrastructure strategy has consistently underinvested in the maintenance and preservation dimension — optimising for creation and sharing while neglecting the longer-term archival challenge.

5. Data as a Public Good Requiring Collective Governance

Perhaps the most consequential theme across all three sources is the call to treat humanitarian data and memory as a public good — collectively financed, governed by shared standards, and maintained for the benefit of the entire sector and the communities it serves. The OCHA report makes this explicit in its calls for collective financing and the UN80 Initiative. The HAE project embeds it in its use of the Internet Archive as a public repository. The toolkit positions itself as a freely available common resource. The challenge is translating this principle into durable institutional architecture.

Climate Data: A Model for Cross-Sector Resilience

Against the prevailing narrative of decline, climate hazard data availability remained stable year-on-year — one of only a handful of sub-categories to hold its ground in 2025. The reason is instructive: climate data is provided not primarily by humanitarian organisations, but by scientific and research institutions — the EU Joint Research Centre, JANUS Atmospheric and Environmental Research, and the UC Santa Barbara Climate Hazards Center — whose funding and mandate lie outside the humanitarian system's current crisis.

The three anchor datasets — ASAP (agricultural anomaly hotspots), FloodScan (near-real-time flood mapping), and CHIRPS (rainfall anomalies) — collectively cover the earth's most food-insecure regions with update frequencies ranging from daily to every five days, and historical baselines stretching back to the 1980s and 1990s. WFP processes CHIRPS data for 129 countries. An NGO in East Africa, Sauti, turns this data into simple SMS messages that help farmers and traders plan in real time.

Key Events in the Humanitarian Knowledge Crisis (2022–2026)

What a Resilient Humanitarian Knowledge System Requires

The convergent analysis across all three sources points to eight actionable priorities. These range from immediate emergency interventions to longer-term structural reforms. Absent political will and collective financing, even the most technically sound toolkit or data platform cannot substitute for the institutional commitment that has been lacking.

Memory as Infrastructure: The Stakes of Getting This Right

Humanitarian operations have always faced the tension between the urgency of the present and the requirements of the future. The current crisis — in which funding cuts are simultaneously degrading real-time data coverage and destroying the institutional memory that would allow the sector to learn from what it has done — makes that tension acute. A sector that cannot track what is happening now and cannot remember what it did before is flying blind in both directions.

The responses documented here — OCHA's sustained management of HDX, the Humanitarian Archive Emergency project's practical toolkit, the cross-sector climate data partnerships — are serious and well-designed. But they are operating at the margins of a structural problem that requires a structural response. The Humanitarian Reset and the UN80 Initiative offer a rare moment of system-wide reflection. The question is whether knowledge infrastructure — data systems, archive preservation, institutional memory — will be treated as the core operational asset it is, or whether it will again be the first line item cut when budgets tighten.