The general debate ends with the people who are missing from the data

What happened

The general debate ended the way the Forum’s debates end, which is to say that it stopped. There was no summing up, and unlike the two sittings before it, no rights of reply. A Vice-President of ECOSOC said “we’ve concluded the general debate of the high-level segment. The meeting is adjourned,” and that was the last sentence of three days and well over a hundred statements.

What makes this sitting worth reading is not the ministers. It is the order of the queue. The debate opened on 13 July across the 33rd and 34th plenaries, resumed at the 35th on 14 July, and finished here, and by the time it finished the room had inverted. Thirty-one member states went first. Seventeen UN and intergovernmental bodies followed. Then twenty-seven major groups, stakeholder groups and accredited NGOs spoke in an unbroken run to the adjournment. Governments had three minutes each. Everyone in that final run had ninety seconds.

And in those ninety-second slots, a genuinely concrete ask surfaced, repeated by group after group who had no obvious reason to coordinate: the SDG statistical system does not contain us, and here is the instrument that would fix it. After three days in which the measurement argument was about whether countries could show their numbers, the last hour was about who is inside the numbers at all.

This page covers the final sitting of the general debate, on the morning of 15 July, and pairs with our coverage of the 33rd, 34th and 35th plenaries. The debate ran under the sub-theme “Delivering better: accelerating urgent and transformative action to achieve the SDGs by 2030.”

Six constituencies, one complaint

Read the closing run for a single variable and the pattern is hard to miss. These are different constituencies with different politics and no shared drafting process, and they made the same structural claim in near-identical terms: they are absent from national statistics, so they are absent from the goals.

• The stakeholder group for communities discriminated on the basis of work and descent, speaking for what it put at 330 million people and naming them (Roma in Europe, Quilombola and Palenque communities in Latin America, Dalits in South Asia, and others in Africa), said that “too often, our communities remain invisible in national statistics and absent from public policies designed to leave no one behind,” and that “our realities remain absent from the national SDG reporting.”
• The LGBTI Stakeholder Group made the mechanism explicit, from housing denial to transgender people avoiding public water points: these realities “remain invisible because the most marginalized populations … are currently not present in national data.”
• The Indigenous Peoples Major Group, co-convened by Tebtebba and the International Indian Treaty Council, asked member states to “strengthen indigenous-led data collection, and ensure data disaggregation by indigenous identity, respecting indigenous data sovereignty and community protocols.”
• The Persons with Disabilities Stakeholder Group asked that partnerships “move beyond consultation to co-creation … backed by adequate financing and data disaggregated by disability.”
• The Major Group for Children and Youth put it as a formal recommendation: “strengthen accountability by complementing national data with citizen-generated, community-based, and disaggregated data.”
• The Saudi Green Building Forum arrived at the same place from the built environment, arguing that the shortfall is not ambition but “persistent gaps in data measurement and accountability”, and that sustainability has to be made verifiable before it can be claimed.

One speaker named the instrument, and it already exists

A demand to be counted is easy to make and easy to absorb. What lifts this above the usual is that one of the last speakers named the specific UN instrument that would do it, and it is not a proposal. It is an adopted framework sitting unused.

REEF Scuba, an ocean conservation NGO with ECOSOC consultative status, used its ninety seconds on the Black Sea and then on this:

The claim checks out. The Copenhagen Framework on Citizen Data was drafted through the UN Statistics Division’s Collaborative on Citizen Data and endorsed by the UN Statistical Commission at its 56th session in March 2025. It supplies exactly what the six constituencies above were asking for: an operational definition of citizen data, principles for producing it, and a roadmap for folding it into official statistics. It is the standing answer to “we are not in the national data.”

It is also, so far, an answer nobody in the closing run reported using. Not one delegation in the three-minute half of this debate mentioned it. The pattern should be familiar to anyone who has followed this Forum: it is the Multidimensional Vulnerability Index problem in a second domain. As the SIDS review and the Beyond GDP session both recorded, the Index was built, adopted, and made binding on no one, and no development bank uses it to decide anything. The citizen data framework is at the same stage, one rung earlier: endorsed, published, and load-bearing for nobody.

The one group that asked for the opposite

The exception is instructive, and it came first in the stakeholder run. The Business and Industry Major Group did not ask to be counted. It asked to be separated.

Read against the six statements that followed it, that is a request from the one constituency in the room that is already visible to the state. Business does not need a disaggregation category or a citizen data protocol to be seen by a finance ministry; it needs a meeting without the others in it. Everyone else in that run was asking for the reverse, to be admitted to a count that currently passes over them. The debate closed with both asks on the record and no mechanism to weigh one against the other.

Meanwhile, the ministers counted what they had built

The state half of the sitting was thinner than the two before it, and it is worth saying so plainly. Where the 34th produced Guatemala on political time and Fiji on missing data, and the 35th produced Myanmar reading its own collapse into the record, this one mostly produced infrastructure. The delegations that brought numbers brought access rates, and they were rising. All figures are as each country reported them about itself.

• Côte d’Ivoire reported electricity access rising from 34% in 2011 to more than 94% today, and said a financing round for its 2026 to 2030 national development plan on 8 and 9 July drew commitments it put at roughly four times the need.
• Togo, presenting itself as one of the most consistent VNR countries with a sixth review, said 1.8 million people have been connected to piped water since 2020, for 86% coverage, and that electricity access reached 75.6% against 50% in 2020.
• Ghana put electricity access above 89%, among the highest in its subregion, and committed to net zero by 2070.
• Sri Lanka reported renewables at around 55% of electricity generation against a 70% target for 2030, and said it will present a third VNR in 2027.
• Trinidad and Tobago was the useful counterweight, reporting renewables at less than 1% of its energy mix and a 30% target for 2030. It also pressed the Multidimensional Vulnerability Index, “to ensure vulnerability is properly reflected in global financing frameworks beyond income measured metrics.”

One measurement detail from this half is worth flagging on its own. Asked to show progress in a UN chamber, two delegations reached not for the UN’s own indicator database but for a league table published by a research network. Slovenia noted that “this year’s sustainable development report ranks Slovenia as 15th among the UN member states.” The Dominican Republic cited “the SDG Index 2026, recently published” as evidence of better water and sanitation. Neither is a UN product. When a country wants a number that sounds like a verdict, the ranking is what is available, which says something about what official monitoring does and does not supply.

The debate about the Declaration finished before the Declaration did

The timing here is the point. This sitting rose at one o’clock. Three hours later, in a different room, ECOSOC took up the draft Ministerial Declaration (E/HLPF/2026/L.1), amended it by recorded vote and adopted it. The Forum’s one negotiated output was still a draft for every single speaker on this page, and the debate that closes the segment closed before the text it is nominally about existed in final form. UN Women was precise about this, welcoming the commitments in “this year’s draft ministerial declaration.”

Almost nobody else engaged with the text at all. The one speaker who had plainly read it line by line, for whether her constituency made it in, was the stakeholder group for communities discriminated on the basis of work and descent. It acknowledged “the inclusion of systemic barrier as a terminology, but we are also slightly disappointed that the” and there the microphone cut out. The sentence that would have said what the draft failed to do is not in the record.

Nobody staged that. It is what a ninety-second limit does to the only speaker doing textual accountability work on the outcome document. And the format offered no route from that sentence to the room where the text was decided: when the Declaration did change that afternoon, it changed because a member state moved an amendment, and what it lost was the COP28 energy commitment from the SDG 7 section, in a year when SDG 7 was under review.

Our read

Also on the floor

• The debate ends in a mailbox. The chair opened by noting that as this was the final meeting of the general debate, any statement not delivered should be “submitted in writing to e-statements@un.org for electronic posting.” The queue outran the clock, and the overflow becomes a web page rather than a spoken record.
• The anchor figure held. Croatia, opening the sitting at deputy prime ministerial level, cited the same score the whole debate ran on: “only 36% of the SDG targets are on the track,” consistent with the Secretary-General’s 2026 progress report. Three sittings, dozens of delegations, no argument about the number.
• International IDEA supplied the sharpest institutional finding of the middle stretch: democratic backsliding is itself an SDG problem, and most of the SDG 16 targets it measures through its Global State of Democracy indices have regressed globally since 2015, with more countries declining than advancing.
• The Volunteers Stakeholder Group gave the run its hardest line, listing volunteers working through war and disaster and then the contradiction: “Volunteers are not the decoration of sustainable development; they are often the reason sustainable development survives it all.” Its microphone also cut mid-sentence.
• The Inter-Parliamentary Union made the case against giving up on arithmetic grounds: “Just because we can’t get the whole job done completely, it doesn’t mean that we should let go of it altogether.”
• The ITU reported that 2.2 billion people are still offline, and pointed back at Geneva Digital Week, the WSIS Forum, the AI for Good summit and the first UN Global Dialogue on AI Governance, all of which we covered. UNIDO noted it co-led this year’s SDG 9 in-depth review with UN DESA.
• Mongolia used its slot to invite the membership to UNCCD COP17, which it hosts in Ulaanbaatar from 17 to 28 August 2026 under the theme “Restoring Land, Restoring Hope.”
• Be honest about the tail. Not all twenty-seven closing statements were of a piece. The run also contained an NGO proposing that everyone on Earth vote on the SDGs by smartphone, a jewellery trade body reporting a reservoir named after itself, and several organisational advertisements. The signal in that run is real, but it is not evenly distributed, and the ninety-second format does nothing to sort one from the other.

Why it matters for the SDGs

This is SDG 17 at its most exposed, and specifically target 17.18, the commitment to high-quality, timely and disaggregated data. The VNR sittings found that coverage is improving fast and disaggregation is not: governments can count more of the framework than ever, and still cannot see their own margins. This sitting is the other side of that finding, delivered by the margins themselves. The people the disaggregation would make visible came to say they are not visible, and they were given half a minister’s time to say it.

The tension the whole three-day debate leaves unresolved is about what a review is for. The format assumes that reporting is done by states, about states, and that everyone else comments afterwards. But the gap the goals now turn on is not a gap in state reporting, it is a gap in what state reporting can see, and the constituencies who can name that gap are the ones the format seats last and times shortest. The instrument to close it has existed since March 2025 and carries the Statistical Commission’s endorsement. On the evidence of this meeting, being endorsed and being used remain two entirely different things.

Watch & read

• The debate’s earlier sittings: the 33rd (where the blocs asked to be measured by vulnerability), the 34th (the “show your work” test), and the 35th (the countries counting their own reversals).
• The Copenhagen Framework on Citizen Data, the instrument the closing run asked for, and the UN Collaborative on Citizen Data that produced it.
• The HLPF closing, three hours later and one room over, where the draft this debate never quite discussed was amended by recorded vote and adopted.
• The 11th meeting, where seven governments reported better SDG data and their own civil society answered that the averages still hide the margins.
• Beyond GDP and the SIDS review, on the Multidimensional Vulnerability Index as the worked example of an adopted metric that changes no decision.
• UN Web TV, ECOSOC, where the recordings of the high-level segment plenaries are posted, and the official programme page for the debate.
• The Sustainable Development Goals Report 2026, the source of the “about a third on track” figure every sitting leaned on.
• Full HLPF 2026 coverage.

Quotations are lightly edited from an automated (Otter.ai) transcript of the UN Web TV recording and should be read as close paraphrase. Statements delivered in Spanish, French and Arabic are summarised rather than quoted, because the automated transcript does not render them reliably. Delegations are cited by country and stakeholders by organisation, as the recording does not render speaker names dependably. Figures are as each country or organisation reported them and were not independently verified, except where a source is linked. Several statements, including two quoted here, end abruptly because the microphone cut at the time limit.