Who Gets to Know? The Grand Bargain's Missing Commitment

The 19–20 October Grand Bargain annual meeting closed its 10-year era and launched its successor, the Collective Action Platform, a 4-year sprint from 2027 to 2030. The independent report commissioned for the anniversary was published alongside it. Most of the commentary since has focused on money and governance: who sits on the Facilitation Group, what happened to the 25% target, whether a shrinking system can reform at all.

This piece examines a less discussed dimension of the report's findings: what 10 years of the Grand Bargain reveal about the production and control of knowledge in the humanitarian system. The Grand Bargain's commitments on transparency and on joint needs assessment asked organisations to publish and share their data; none gave local actors or affected people authority over the evidence itself: who produces it, who holds it and who can contest it. That authority is the missing commitment of the title.

The evaluators documented a paradox. The number of signatories meeting the 25% localisation funding target rose from 13 in 2022 to 23 in 2025. Over the same period, the share of humanitarian funding actually reaching local and national actors fell from 10.8% to 8.7%. The two measures count different things: the first counts signatories, many of them small agencies and NGOs, while the second weights every dollar, so the largest donors and agencies decide the share.

More organisations hit the target; less money arrived. The same pattern repeats on gender: self-reported funding information on women-led organisations increased by 450%, while the funding itself dropped by 30%. Visibility increased while funding declined.

The two patterns share a cause: a system that measures what it reports about itself. For 10 years the Grand Bargain looked for policies, strategies, reporting rates and self-assessments, and that is what it found, in abundance.

The independent report is clear about the result: self-reporting reached 90% compliance and worked “more effectively as a transparency and learning mechanism than as an accountability tool”. No focus area reached substantial progress. Movement happened in representation, policy and process, while resources, authority and risk stayed where they were.

ODI's comparison of 5 UN intermediaries' Grand Bargain self-reports with what its Humanitarian Policy Group could trace in independently verifiable datasets shows how far the two diverge. In 2024, WFP reported US$1,815 million in direct funding to local and national actors and the Humanitarian Policy Group traced US$261 million; for OCHA's country-based pooled funds, the figures were US$329 million and US$334 million.

The metrics failed to register this earlier because the knowledge architecture beneath them was never localised, and was barely functional. Two-thirds of trackable humanitarian funding cannot be followed beyond the first recipient. The three systems that are supposed to show where the money goes, the Financial Tracking Service, IATI and the Grand Bargain self-reports, are inconsistent enough that the report concludes a reliable assessment of funding reaching local actors is not currently possible. Ten years into a reform whose flagship indicator is a funding share, the system cannot reliably compute that share.

ODI's series on funding to local and national actors, which PRISM reproduces on its localisation page, shows how far the share depends on what is counted. With the funding that UN agencies pass on to local partners, published only from 2022, the share was 6.7% in 2022, 10.8% in 2023 and 9.5% in 2024. Without it, the share was 3.5%, 6.8% and 5.2%. Counting only funding given directly, the measure closest to the target's own wording, “as directly as possible”, the share was 2.3%, 4.6% and 3.8%. Part of the rise in the measured share after 2021 came from UN partner data becoming visible.

The same architecture governs how needs are defined. The report finds that engagement with affected communities “still tends to stop at information-gathering rather than shaping strategy, funding and programming”. ALNAP's 2026 State of the Humanitarian System report found, in parallel, that only 13.6% of practitioners said a new needs assessment drove prioritisation under the funding cuts: assessed need played a small part in deciding what was funded.

I would put it less diplomatically: communities are treated as data subjects. They are surveyed when an assessment needs filling, through questions designed around what the agency already intends to propose, often sampled from beneficiary lists and phone surveys that miss people without a registered phone. The information flows one way, upward, into proposals and reports the community will never see and did not shape. Accountability, in practice, runs towards donors; it reaches communities last, if at all.

The Grand Bargain's commitments concerned decisions as much as money, and decisions run on knowledge: who defines need, who counts, and whose evidence is believed when accounts differ. A local organisation can receive a sub-grant and still have no say over the assessment that framed the project, no access to the funding data that would let it contest its share, and no standing to challenge an intermediary's numbers. Localisation so far has moved the labour and left the knowledge where it was.

Knowledge also moves downward, as definitions: partners build their proposals, theories of change and sectoral focus around the priorities a donor sets in its funding texts, so the text decides what partners can frame, propose and report. In PRISM's reading of DG ECHO's Humanitarian Implementation Plans for Palestine, mentions of the nexus fell from 11 in the 2023 decision to 1 in 2026, and Area C left the decision text entirely.

The report names the consequence under funding cuts: localisation in reverse. Funding to local and national actors fell by 27% in 2025, while they took on more frontline responsibility and more risk. They carry more of the work and less of the power to describe it.

The new framework contains one change of a different order. It phases out self-reporting in favour of existing data systems and, for the first time, “ground-truth evidence generated directly by National Reference Groups”, which also gain 2 voting seats and a dedicated funding window. This is a significant admission: 10 years of self-reporting could not establish what the reform achieved, and the evidence is now expected to come from those closest to the response.

At the annual meeting, Liudmyla Leonova, co-chair of Ukraine's National Reference Group, told the signatories to stop treating local actors as a consulting service. Ground-truth evidence is the alternative to being a consulting service: it makes local actors the authors of the record that others cite.

Whether it becomes that depends on three conditions the next four years will test. Evidence production needs money, because monitoring is paid work; the same report documents what happens to unfunded mandates. It needs methods authority, meaning National Reference Groups define what is measured and how, and design the questionnaires they administer.

And it needs standing: when ground-truth evidence contradicts an agency's self-report, the contradiction has to be answered. If any of the three is missing, ground-truth evidence will join the list of commitments the sector reports on and leaves unmet. PRISM will report on all three each year until 2030.

Evidence matters for what it makes possible. Systematic monitoring and data collection are the basis of humanitarian diplomacy, of negotiations for access and of the implementation of international humanitarian law; whoever keeps the record holds the material those processes run on.

This is why I built PRISM. One person with open data can now check numbers that used to require an institution: funding flows, protection records, the vocabulary of financing decisions, across more than 230 countries. I built it because the evidence base of humanitarian action should be open to anyone who wants to check it, and because every independent pair of eyes on the sector's numbers changes, in some measure, who can hold whom to account.

In June 2026 PRISM stopped showing the Grand Bargain self-reports and moved its localisation pages to ODI's funding data, the change the Collective Action Platform now makes. PRISM's registers follow the principle of ground-truth evidence: analysts file entries under their own name or a label they choose, an analyst's entry takes precedence over an aggregated record citing the same report, and every entry has a permanent link and can be exported, so the people closest to an event are the authors of its record. The platform's contributor desk is open to local researchers and practitioners for the same reason.

Humanitarian financing is commonly described as needs-based. The Grand Bargain's first decade suggests that funding follows knowledge, and that knowledge follows power. The new platform is the first iteration of this reform to acknowledge this in its design. The test for 2030 therefore covers more than a percentage: whether the people the system exists for hold the power to know, the power to act upon what they know, and the power to deliver with quality. The other two depend on the first.