Summary

  • The OECD received 42 consolidated government responses on central or federal practices as of 1 January 2026.
  • 26 of 40 respondents reported reducing internal rules as a significant measure; 30 of 40 named task automation, including AI. Those are reported measures, not measured service results.
  • The OECD calls its proposed indicator framework a shared diagnostic language, not a country ranking. It says countries still need to agree objectives, map data and build comparable measures.

Analysis

A government can report a new team, a rules review or an automation project in one line. A resident faces a different test: how long did the service take and how many steps did it demand? The OECD’s 2026 Government Unstuck report has 42 consolidated responses about reform activity. It proposes a second set of measures for what users experience.

The survey was administered from February to June 2026 and sent to all 38 OECD members and eight accession candidates. Forty-two responses came back: 35 from members and seven from candidates. National officials coordinated a single whole-of-government response on central or federal policies, practices and arrangements in place on 1 January. The OECD then ran validation and verification rounds with respondents.

That process gives the report a useful map of what administrations say they are doing. It also defines what the map cannot answer. Twenty-six of 40 respondents identify reducing internal regulations and procedural burden as a significant efficiency measure. Thirty of 40 say task automation, including AI, is central to their efforts. These are counts of reported measures. They do not say how much time a permit now takes, whether an internal approval disappeared for good, or whether a service user had to submit the same information twice.

The distinction matters because a government can count its own activity faster than a resident can feel its effects. A new team, a process review or an automation project creates a record inside the administration. The public result is a different record: request-to-delivery time, repeat contacts, the number of steps, whether an application succeeds without escalation, or whether feedback changes the next version of a service.

The report calls the space between reform inputs and long-term impacts the “missing middle.” It proposes four directions for measuring that space: a renewed relationship with citizens; simpler government and regulation; more room for manoeuvre in public finances; and a shift from prepared to proactive government. The annex makes those aims concrete. It suggests, among other examples, measuring average time from a service request to delivery, the number of steps and contacts needed for a life-event transaction, and the share of consultations followed by a published government response.

Those are examples for a framework under development. The report says the framework is not a composite index that ranks countries. Its purpose is a common diagnostic language for identifying whether reforms are moving in the intended direction. The OECD proposes a phased process: settle the intermediate outcomes, map the indicators that already exist, identify gaps and then build measures with countries. Some data could be collected directly by the OECD; other indicators may depend on national systems using shared definitions. The report leaves that balance open for discussion.

The report puts two counts side by side: 29 of 31 OECD countries collect user-experience data; 16 of 31 say it informs service-improvement decisions. This is a separate evidence base from the 42 responses to Government Unstuck. Collecting feedback creates a record. A decision log shows whether that record changed a service.

The separate 2026 OECD Trust Survey offers a citizen view of the relationship, not a direct measure of the effect of any one reform. It covers 33 OECD members and five accession candidates. Across OECD respondents, 40% report high or moderately high trust in national government, against 43% with low or no trust. Thirty-one percent say the political system lets people like them have a say; 39% say taking part in a public consultation influences government action. Those results show why the OECD treats experience and voice as part of government performance.

The survey does not link respondents to a specific automated service or compare their answers before and after a reform.

A useful measurement system will join four kinds of evidence the report identifies: what people experience, what administrative records show, what users and officials do, and what structural or fiscal data reveal. Each answers a different question. Perceptions can move after events unrelated to a service. Administrative records can count completed transactions while missing people who gave up. Behavior can change for reasons outside government. Cross-country records need shared definitions before one country’s “processing time” can be compared with another’s.

That is a governance choice as much as a statistics project. The report says member countries will help define the outcomes and methods. The OECD may collect some measures itself or convene national systems around common definitions. Whoever gathers the data should publish the denominator, the service boundary and the rules for revision. Otherwise governments can compare labels while counting different things.

The OECD should publish an owner, denominator and revision rule for every service indicator. It should pair each reform claim with the wait, number of steps, repeat requests or consultation response it can count. A public series like that would let residents see whether government activity changed the service they use. Make that the next report’s first page.

Sources