This chapter examines the capabilities and practices that can make the transformation to a “Government Unstuck” possible. Drawing on experience across OECD and partner countries, it focuses on four interconnected areas: trusted and connected digital technologies and data; skilled and empowered public servants; agile government operating systems; and meaningful and strategic citizen engagement. For each, the chapter identifies the challenges holding governments back, highlights defining practices that are enabling progress, and considers the capabilities required for the future. Together, these practices show how governments can build on what is already working to accelerate change and strengthen democratic governance.
Government Unstuck
The Next Frontier of Government Transformation
Abstract
Chapter 2 set out the direction that governments need to move in to become more trusted, effective and resilient. Chapter 3 outlines the tools and capabilities that allow that change to happen and that, ultimately, make the transformation to a “Government Unstuck” possible. Evidence that this direction is already taking hold can be found in the practices that governments across OECD and partner countries are building, testing and bringing to scale. Chapter 3 of this report therefore turns to four families of practices that provide evidence that countries have begun moving towards a Government Unstuck:
1. Trusted and connected digital technologies and data, which enable governments to act more intelligently, proactively and responsively;
2. Skilled and empowered public servants, who provide the leadership, capabilities and culture needed to sustain change;
3. Agile government operating systems, including centres of government, regulation, public procurement and service design, which translate ambition into co-ordinated action; and
4. Meaningful and strategic citizen engagement, which ensures that day-to-day governing and reform is informed by societal perspectives, strengthened through participation and grounded in public legitimacy.
The four families listed above each encompass a wide variety of practices, which can be grouped into distinct areas. The table below provides a synthesis of these, to ease navigation of this chapter.
Families and areas of practice
For each of these families of practices, the report examines the challenges that continue to hold many governments back, highlights the practices from across OECD and partner countries that are proving decisive in getting government unstuck, as well as explores the forward looking tools and capabilities that will be needed in the years ahead. Rather than offering a fixed blueprint, it identifies where the direction of change is already becoming concrete and where governments can build on what is working to accelerate progress. Taken together, the direction of change and the practices that carry it forward form the Government Unstuck agenda: a practical framework for helping governments move beyond today's constraints and build stronger democracies for the future.
3.1.1. Introduction
Governments across the OECD are drawing on digital technologies and data in every part of their work. Digital identity and data-sharing systems, rules for governing data, the responsible use of AI, and the management of digital investments and dependencies are being built, tested and brought to scale. The pattern of that activity shows countries moving in the direction of change.
Yet much also remains to be done: the OECD Digital Government Outlook 2026 finds that, while most OECD countries have the right governance and strategies in place, implementation, and thus, actual positive impact for people and businesses, remains a challenge (OECD, 2026[1]).
These areas of practice reinforce one another. Sound data lets governments understand what people need, act ahead of problems and adjust as circumstances change. Shared digital public infrastructure gives governments tools they can reuse rather than rebuild from scratch. AI extends both where it serves a defined operational purpose rather than because it is available. And responsible governance is what allows all three to operate at scale with public confidence. This section tackles these three key areas in that order and highlights defining practices for each (see Table 3.2).
Defining practices for Trusted and connected digital technologies
3.1.2. Advanced data foundations and digital infrastructure
Key issues
The OECD 2026 Digital Government Outlook shows that countries have made real progress in building advanced digital and data foundations in government. Almost all OECD countries have a national data strategy, most have formally adopted the once-only principle, and interoperability systems that allow data to be exchanged or linked across government are in place in 30 of 36 countries (OECD, 2026[1]). Digital identity, single digital gateways, digital post and payments, and shared cloud services are also widely available, with 74% of core digital public infrastructure components present on average (OECD, 2026[1]). These foundations make it easier for different parts of government to work together, respond rapidly and operate more efficiently.
The challenge now for countries is implementation, that is, to ensure that the benefits of these digital foundations become tangible for people and businesses in their interactions with government. Indeed, digital identity is often still deployed as a stand-alone authentication step rather than as the layer through which services are joined up. A plethora of data are available to be shared, but the day-to-day practices needed to keep data accurate, interoperable and safe to share have not kept pace. For instance, only 63% of public institutions on average share data through their national interoperability system (OECD, 2026[1]). While 33 of 36 countries have a national digital identity strategy, only 25 report that more than half of their online services are accessible through a secure digital identity (OECD, 2026[1]). The challenge is even more acute regarding sensitive administrative data, such as that on earnings, benefit status, civil registration records, contributions, which agencies hold under different rules and cannot readily share with each other. The result is that people still experience government as a set of separate parts: repeating information government already holds, proving who they are more than once, and pursuing entitlements service by service. This is the distance still to travel to fulfil a simple government as described in Chapter 2.
Implementation lags partly because nothing or no one compels it. Most countries have now assigned an institution responsibility for overseeing digital and ICT investment priorities, but a mandate on its own is not enough to drive significant transformations. Unless that institution sets common standards, requires data reuse and reviews major investments before funding is committed, agencies keep building their own versions of what already exists and the shared platforms that do exist degrade without the continued funding and attention needed to keep them working.
Implementation can also be uneven when it depends on technology governments do not build or control themselves. Governments depend heavily on outside providers for the technology underpinning these systems, which is normal and unavoidable. Yet, this reliance can become a risk when it is not carefully managed: when governments have no clear picture of which providers, technologies and jurisdictions their critical services rest on; when they lack the in-house expertise to specify, supervise and if necessary, replace what they buy; or when too much rests on a single supplier or country with no practical route to switch. The question for countries is therefore which capabilities a government must retain to remain able to change course and switch providers in case a service fails or no longer meets the standards. That ability to switch is itself a margin of manoeuvre, and a condition of a prepared government as outlined in Chapter 2. A government that cannot move has less room in its budget, and no answer if a provider fails.
Finally, everything that makes implementation hard within a country is harder across borders. Interoperability between countries depends not only on technical connections but on mutual recognition of systems, clear liability arrangements and harmonised protections for privacy and security, conditions that remain largely unresolved. Only 68% of OECD countries include cross-border service goals in their data strategies (OECD, 2026[1]). For people and businesses moving between countries, the effect is the domestic problem repeated at a larger scale: paper, in-person steps and the same information provided over again.
In sum, becoming “unstuck” now depends less on what countries have built than on what they do with it. A growing number of countries are showing what this takes in practice.
Defining practices
Making the once-only principle operational. The gap between formal adoption, which is widespread across OECD countries, and working reality remains significant. Countries that have made the once-only principle a reality for business and citizens have managed to combine three necessary elements: legal clarity on what data can be shared across agencies; technical systems capable of connecting with each other; and a binding requirement that whenever a new service is designed, it should avoid asking citizens and businesses for information government already holds. Belgium’s Only Once Act and the EU Single Digital Gateway, including its Once-Only Technical System (OOTS), provide examples of leading practices, with the latter setting a cross-border benchmark for how such obligations can be standardised across jurisdictions.
Unlocking data for sharing and reuse. Governments collect and generate vast quantities of data, but much of it remains locked within the institutions that generated it. Closing this gap requires interoperable data infrastructure that allows institutions to share data across organisational boundaries, and embedding its use across government functions, not just by central policy teams. Italy’s National Digital Data Platform (PDND) illustrates how a national interoperability infrastructure can shift from a model of data publication to one of active data use, allowing agencies to share and access data in real time.
Enhancing data quality and accuracy. High-quality data is essential for reliable sharing and reuse across government and AI systems. Data quality frameworks can guide institutions that collect, generate and maintain data, while common semantic standards ensure that different system interpret exchanges data consistently. Without it, interoperable systems exchange unreliable data, proactive services act on inaccurate information, and AI systems reproduce and amplify existing biases, making data governance a prerequisite for responsible deployment. Sweden’s national data portal catalogues specifications, including information models, schemas, profiles, ontologies and related terminologies, to guide the publication of comparable data and improve interoperability across datasets.
Building the foundations for cross-border data sharing. Technical interoperability is a necessary starting point, but cross-border data sharing ultimately depends on a shared foundation of trust, mutual recognition of systems, clear liability frameworks and harmonised protections for privacy, security and personal data. The EU Single Digital Gateway provides a regional model; beyond it, progress has been limited and largely bilateral. The European Commission’s Public Procurement Data Space illustrates how shared data infrastructure at regional level can generate transparency and comparability that individual countries could not achieve alone.
Getting the most from shared digital public infrastructure. Countries increasingly manage digital identity, payments, service gateways, and notifications as a coherent portfolio of tools, rather than as disconnected projects. Shared components need common standards and interfaces, concerted governance arrangements including sustainable funding and political backing, and a presumption of reuse backed by a transparent process for justified exceptions. Success should be measured by adoption, reliability and benefits realised, not by whether a platform technically exists. Denmark's MitID shows one way to secure that long-term backing: a public-private partnership between the Agency for Digital Government, regions, municipalities and Finance Denmark provides secure shared access across government, banking and private digital services nationwide, spreading both the cost and the institutional stake in keeping the system working.
Making digital identity a trusted open gateway to services. Countries have found that treating digital identity as an enabling layer for joined-up services, verified credentials and secure transactions, rather than as a stand-alone authentication mechanism, changes how services are designed. The remaining gap is adoption, including among service providers and people who face barriers of trust, usability, accessibility or connectivity. Common approaches have combined technical standards and clear liability rules with privacy-by-design, user control over what is shared, strong fraud monitoring and meaningful non-digital alternatives. At the European level, the EU Digital Identity Wallet shows how common specifications can support interoperable digital identity infrastructure, enabling verified credentials to be used securely across borders and across public and private services.
Building capabilities where it matters. Effective digital government requires a central or federated authority with a clear mandate to set standards, co-ordinate institutions, develop shared components and oversee major investments throughout their lifecycle. Australia’s Digital and ICT Investment Oversight Framework provides an integrated whole-of-government approach to managing digital and ICT investments across the entire project lifecycle, from planning and prioritisation through sourcing, assurance and benefits tracking (OECD, 2025[2]). Effective oversight also depends on the skills and resources to manage suppliers and operate critical systems. In Germany, the State of Schleswig-Holstein migrated to open-source alternatives by investing in in-house technical capacity and engagement with open-source communities, reporting annual licence savings of €15 million against a one-off transition cost of €9 million (State Chancellery of Schleswig-Holstein, 2024[3]). Korea’s specialised digital service contracting system takes a complementary route through procurement: small and medium-sized enterprises accounted for 75.8% of selections by February 2026, widening the domestic supplier base while reducing concentration in large international incumbents (OECD, 2026[1]).
Forward looking issues
The practices set out above show that a growing number of governments are laying credible data and digital foundations and treating shared infrastructure as a governance question. Yet the frontier remains open: availability is easier to build than active use, mandates are easier to assign than to make consequential, and cross-border sharing raises questions of trust, liability and mutual recognition that no jurisdiction has yet resolved. Three areas stand out as priorities for further exploration, both for data and digital foundations themselves and for the OECD's future work in supporting them:
Measuring effect rather than existence. The central challenge is translating existing building blocks, such as data strategies, interoperability systems and data governance frameworks, into tangible impact on services, policy and institutional performance. Current evidence captures data availability reasonably well, but not whether data is actively used to improve outcomes, or whether oversight delivers. Closing this gap requires examining which data governance arrangements deliver public value in practice across different government functions, and extending measurement frameworks, including the Digital Government Index (DGI) and OURdata Index, to track the impact of rather than availability, and identify the institutional conditions that make mandates consequential rather than nominal.
Identifying when dependency becomes a risk. Governments need shared metrics for what counts as excessive dependency concentration, a threshold at which reliance on a given provider, technology or jurisdiction crosses from ordinary efficient procurement into strategic risk for public-service continuity. The measurement infrastructure is starting to emerge, including through the OECD's work on tracking domestic public cloud compute availability for AI, but the governance layer on top remains missing: a shared basis for deciding when concentration becomes a risk, and clarity on who within government is responsible for making that call.
Building the governance conditions for cross-border sharing. The OECD's comparative reach is well placed to identify converging models of mutual recognition and liability-sharing across different legal traditions, and to extend lessons beyond existing regional frameworks. The priority is not more diagnosis of the problem but clearer comparative evidence on the governance conditions - legal, technical and institutional - under which data sharing actually delivers at scale, and practical guidance governments can act on. The EU’s Single Digital Gateway and Digital Identity Wallet provide concrete examples at scale of how countries can promote digital access to services across borders and sectors (see Box 3.1).
EU’s Single Digital Gateway
The EU’s Single Digital Gateway (SDG) demonstrates how the once-only principle can be translated from an administrative ambition into an operational reality across jurisdictions. Established to help citizens and businesses access information, complete administrative procedures and obtain assistance when exercising their rights across the EU, the SDG is supported by the Once-Only Technical System (OOTS), which enables public authorities to securely exchange evidence and documents across borders rather than requiring users to resubmit information already held by another administration. What distinguishes the EU approach is that it combines technical interoperability with a common legal framework, mutual recognition mechanisms, explicit user consent requirements, and shared standards for data protection and security.
EU’s Digital Identity Wallet
The EU Digital Identity Wallet (EUDI Wallet) extends this logic by positioning digital identity not merely as an authentication mechanism, but as an enabling infrastructure for joined-up services, verified credentials and secure transactions. The wallet allows individuals and businesses to identify themselves, store and share verified attributes and documents, and access both public and private services across borders and sectors while retaining control over their personal data. By embedding digital identity within a broader ecosystem of reusable credentials, electronic signatures and trusted exchanges, the wallet supports a shift in service design: rather than repeatedly asking users to prove who they are or submit supporting documents, services can be built around trusted, user-controlled digital credentials.
3.1.3. AI in government
Key issues
Governments no longer need convincing that AI can create public value. The challenge now lies in making it part of ordinary operations in ways that improve productivity and responsiveness, whilst preserving accountability. What is at stake reaches beyond AI itself: the proactive services, simpler administration and restored fiscal room the direction of change describes increasingly depend on it. AI is now used in at least one government activity in all OECD countries, with its use being strongest for internal processes and public service delivery, but more limited in policymaking and oversight (OECD, 2026[1]) (see Figure 3.1). Ambition is high and likely to continue to rise, yet most of it remains unrealised.
Percentage of OECD countries using AI in the public sector by policy area, 2023 and 2025
Note: Oversight and accountability in the public sector was not measured in the 2023 survey, therefore no comparison is available. Accession candidate countries report the following use rates: Internal processes (25%); Public services (38%); Policymaking (0%); Oversight and accountability (17%). 2025 data not available for Germany and the United States. 2023 data not available for Bulgaria, Germany, Greece, Indonesia, Slovak Republic, Switzerland, Thailand and the United States. Data for Indonesia and Thailand cover 1 January 2022 to 31 December 2023, while Oversight and accountability does not include data for these countries. Refer to Annex Table 4.A.1 of Chapter 4 for comprehensive OECD and Accession country data.
Source: OECD (2026[1]), Digital Government Outlook 2026: From Foundations to Transformational Impact, OECD Publishing, Paris, https://doi.org/10.1787/0496b2bc-en.
Part of the difficulty for governments is sequencing: many initiatives start from an available technology rather than a clearly defined operational problem, making it harder to judge afterwards whether AI was the right tool or whether the right conditions for it were ever in place. The binding constraint is rarely the technology itself, but the governance and enabling conditions around it: quality data, infrastructure, skills, sustainable funding, institutional readiness, and the capacity to use public procurement strategically (OECD, 2025[6]). Projects that start from the technology tend to be one-off tools built on undocumented data and weak systems, so they leave nothing behind and the next initiative starts from scratch.
Capability gaps inside the public administration often compound this issue. Only 36% of OECD countries offer training on using AI in public services or policymaking, leaving many administrations without enough in-house understanding to specify their needs, supervise the tools that they deploy or judge the cost-benefits of the solutions they are being sold. Heavy reliance on external contractors risks a dependency cycle in which governments cannot evaluate or replace what they buy (OECD, 2025[6]).
Even well-founded pilots stall without the machinery to scale them. Only 58% of OECD countries provide central support for procuring AI goods and services (OECD, 2026[1]), and existing procurement systems are not built for technology that is iterative, fast-moving and difficult to evaluate in advance. Without shared mechanisms to fund, buy, adapt, evaluate and, where necessary, discontinue AI initiatives, every institution has to solve these problems for itself.
The evolving frontier raises the stakes further. Governments can draw on a spectrum of AI approaches, from data analytics, rules-based automation and machine learning to generative AI and, at the frontier, agentic AI systems that can plan, use tools and act within set permissions with little human intervention at each step (OECD, 2025[6]). The most capable option is not always the right one for a given problem, and the more a system acts on its own, the more its scope, permissions and reversibility have to be settled before it is deployed rather than after. Extending autonomy in steps, with narrow permissions and the ability to pause or reverse what a system does, therefore becomes the condition for deploying it at all.
Yet a growing number of countries are working through these constraints, and the practices they are building offer the clearest evidence of what it takes to move from ambition to operation.
Defining practices
Starting from the problem, not the technology. Countries that have moved beyond pilots typically begin from the operational problem to be solved – whether prediction, classification and routing, drafting and synthesis, anomaly detection, or cross-agency co-ordination – and ask whether better data sharing, rules-based automation or process redesign would serve as well before reaching for AI. Korea’s e-RFP Assistance System, which combines several AI techniques to draft procurement requests and check compliance, reduced document preparation time by 70% while reporting 99.9% compliance with procurement rules (OECD, 2026[1]). Chile used an AI-enabled management platform to track responsibilities and deadlines across its complex environmental permitting system, cutting overdue agency responses from 33% to 3% in four months simply by making delays visible to all stakeholders and without automating permitting decisions. Japan's Ministry of Internal Affairs and Communications (MIC) has applied automation to policy formation itself, by deploying AI agents for ancillary tasks such as scheduling, bookings and documentation, while freeing policymakers for higher-value strategic work. These examples show that targeting concrete operational bottlenecks first has proved a valuable starting point.
Turning AI projects into reusable foundations. The most durable AI initiatives leave reusable assets behind, through better-documented data and service maps, model evaluation routines, feedback loops, procurement templates, rather than sitting on top of weak systems as one-off tools. Building Portugal’s gov.pt virtual assistant, for instance, meant consolidating dispersed public-service information into a single curated knowledge base; the conversation logs it generates are analysed to find gaps and improve content. This approach is also the most relevant to agentic AI, since systems that act depend on the data they can access, the systems they can reach and the public servants who supervise them.
Building the in-house capability to direct, supervise and adapt AI. Leaning heavily on external contractors risks a dependency cycle that leaves governments unable to specify, evaluate or replace what they buy (OECD, 2025[6]). Governments need not internalise every technical skill, but they need enough to be informed buyers, credible supervisors and active shapers of AI-enabled change. That means role-specific capability: basic AI literacy for all public servants and strategic understanding for leaders, through to domain-specific skills for service, procurement, legal and evaluation teams. Estonia's ABC of AI e-learning module, run by the Digital State Academy, illustrates this tiering at the foundational level; Ireland's AI Masterclass for Senior Leaders targets the strategic judgement leadership roles require, rather than technical depth (OECD, 2026[7]). Portugal’s AI Experience Lab takes a third route: taking public bodies through ideation, certified training and deployment so that teams build capability by delivering real projects. Colombia's CONPES 4144 (2025[8]) National AI Policy sets out a comparable national roadmap, embedding ethics and governance components alongside capability-building measures across the public administration. Chile's VIGIA case illustrates how central support can build agency capability rather than create long-term dependence on a supplier: the Superintendence of Social Security (SUSESO) developed the tool with funding and implementation support from the Ministry of Finance's State Modernisation Secretariat, and while external specialists supported development, SUSESO received the code, documentation and training needed to operate and improve the tool internally. Demographic change offers an important lever: many OECD governments face significant retirement waves over the coming decade and ensuring that recruitment and induction are deliberately designed to attract digitally and AI-literate staff can build capability through generational renewal, not only through retraining. Leaning heavily on external contractors risks a dependency cycle that leaves governments unable to specify, evaluate or replace what they buy (OECD, 2025[6]). The following section on “Skilled and empowered public servants” describes efforts that some countries have undertaken to equip themselves with AI capabilities in-house.
Using procurement and partnerships as scaling infrastructure. Pilots will not scale without mechanisms to fund, buy, adapt, evaluate and, when necessary, terminate AI initiatives responsibly. Korea’s Digital Service Specialised Contracting System replaced lengthy tender processes with catalogue and direct contracting, reducing the time to acquire digital and AI services from months to approximately two weeks. France’s Albert gives public teams shared, secure access to a range of AI models - including open-weight models and procured commercial models - through a single national platform, widening access to innovation while keeping capability and control inside government. Formal procurement access alone does not bridge the gap between policy intent to working deployment: the AI Playbook for the UK Government codifies shared principles and practical guidance on how to select, deploy and govern AI safely, so teams do not start from scratch, while Australia's AI Delivery and Enablement (AIDE) function works alongside agencies to remove process barriers, share reusable solutions and translate what procurement can buy into what teams can actually use.
Creating bounded pathways for agentic AI. Bounded deployment means constraining an agentic system's scope, autonomy and permissions until it has demonstrated reliability: narrow tasks, limited data access, and a human able to review or reverse what it does, rather than granting broad autonomy upfront and relying on oversight to catch problems after the fact. Countries moving toward agentic AI have typically started with specific tasks under close control – agents that only read information rather than change it, that draft work for an official to review, or that support staff rather than act directly on the public – before expanding scope as confidence and supporting systems grow. Australia’s agentic AI addendum to its public-sector AI technical standard sets out controls of this kind: agents act within assigned permissions, are granted only the minimum access necessary for their task, and their activity is observable, auditable and capable of being rolled back. Estonia’s Bürokratt provides an interoperable network of public-sector AI assistants giving citizens one conversational entry point to government, routing each query to the right agency’s assistant behind the scenes, with a fall-back to human staff when needed. Gartner projects that over 40% of agentic AI projects will be cancelled by end of 2027 due to rising costs, unclear value, integration challenges and insufficient governance controls (Gartner, 2025[9]) – a reminder that the discipline of bounded deployment matters. Bounded pathways are only actionable if they generate learning: measuring what bounded AI deployments actually produce remains underdeveloped across OECD countries with only 28% reporting any assessment of their AI use cases – though the UK's guidance on impact evaluation of AI interventions (updated May 2026) and OECD forthcoming work point toward what proportionate evaluation can look like in practice.
Forward looking issues
The practices set out above show that a growing number of governments are moving AI beyond pilots. Yet the frontier remains open: adoption is easier to count than to assess, and agentic systems are arriving faster than the arrangements needed to govern them. Two areas stand out as priorities for further exploration, both for AI in government itself and for the OECD's future work in supporting it:
Comparable, proportionate methods for measuring AI impact in public services: moving beyond counting use cases toward assessing whether AI actually improves productivity, quality of service or equity of outcomes. The OECD is well placed to develop this methodology, building on existing work on public policy evaluation and impact measurement of broader digital investments.
A practical route into agentic AI ahead of wide deployment. The OECD forthcoming work on Governing with Agentic AI will offer governments a framework for understanding agentic readiness, assessing adoption maturity, and identifying pathways for piloting, scaling and, where necessary, pausing systems that act on government's behalf (see Box 3.2). The discipline of bounded deployment – narrow scope, limited permissions, observable actions and retained human accountability – is what separates agentic AI that builds public trust from agentic AI that erodes it. Establishing this discipline before wide deployment, rather than in response to failures after the fact, is the most important near-term priority for governments and for the OECD's support work in this area.
Most governments are still weighing whether, when and how to use AI agents and agentic AI systems (see also Box 2.2 of Chapter 2) to carry out actions on their behalf. The appeal is clear, but so is the difficulty. As these systems take on longer tasks with less step-by-step human involvement, questions of legal authority, accountability and human oversight become harder to resolve, and the responsibility governments carry to protect citizens and uphold due process leaves little margin for error. While government interest is high, most lack clear basis for judging their own readiness and proceeding on sound footing.
The OECD's forthcoming Governing with Agentic AI series is intended to provide one. Building on the Governing with AI report and developed with a working group of Member countries leading on the topic, it will offer a shared understanding of what agentic AI means for the public sector alongside practical guidance on the institutional, legal and technical conditions that need to be in place for trustworthy use. Grounded in the experience of early adopters, the work is designed to support countries across the spectrum of digital maturity and to help each move from general interest toward well-informed, responsible decisions about adoption.
3.1.4. Ensuring trustworthy data governance and AI
Key issues
Accountability is what allows data-driven, anticipatory and AI-enabled government to operate legitimately at scale, rather than a constraint upon it. As governments use data and AI to personalise services, anticipate needs, detect risks or act before a citizen has made a request, public authority can become faster but less visible. Without clear authority, transparent records, human accountability and routes to contest decisions, a more capable government risks widening the accountability gap that erodes trust. This matters especially because only around one in three citizens is confident that government will protect personal information when using AI, maintain adequate human oversight, or be transparent about how AI systems work (OECD, 2026[10]). Building this trust in how AI is used will be essential for the renewed relationship between government and citizens that the direction of change describes.
The problem is partly structural. Many legal, administrative and oversight frameworks were built for a reactive state, in which a citizen applies for a service, an official decides and the reasons can be inspected after the fact. Proactive services and agentic AI reverse that sequence, often without a clear legal basis, and they depend on combining personal data held under different rules, which can have consequences on how privacy commitments are applied in practice (OECD, 2026[10]).
Commitments to trustworthy AI have likewise outpaced the controls that would make them binding. Only 11 of 36 OECD countries (31%) have a formal algorithmic transparency standard and only 14 of 36 countries require an ex ante risk assessment before an AI system is deployed (OECD, 2026[1]). Agentic systems sharpen this, because the question shifts from what a system says to what it does: oversight designed to inspect decisions after the fact offers little once an action has taken effect. Governance is outpaced by change itself, with effects crossing institutional and national boundaries faster than structures adapt.
Yet a growing number of countries are closing these gaps, and the practices they are building show what accountable data and AI use looks like in operation.
Defining practices
Giving proactive government action a clear legal basis. Countries that have moved farthest in proactive service delivery have done so by specifying when and how the state may combine data, work out what someone needs, and act on their behalf before they ask. This means having a law or clear policy that says what data can be used, for what purposes, for how long, with what safeguards and who checks that it’s being used properly. That law should treat low-risk actions, like pre-filling forms or sending eligibility reminders, differently from higher-stakes ones, like managing payments, changing records or initiating enforcement. It should also say when the state needs consent, has to notify the person, needs a human to confirm the decision, or must let people opt out. Austria shows what this looks like for a high-stakes case: since 2015, the tax authority has been able to pay family allowance automatically when a child is born, using data from the civil register, with no application needed (Government of Austria, 2015[11]).
Protecting privacy and modernising data governance. Responsible governance means applying well-established privacy rules to anticipatory government: only collecting, sharing and combining the data actually needed for a clear purpose; building privacy in from the start rather than adding it later; assessing the privacy risks of higher-risk data uses before they begin; and giving people clear rights to know what data is held about them, see it, correct it, and object to its use. Norway's data protection authority runs a sandbox where public bodies can work out whether an AI system has a lawful basis before launching it, whether it's designed with privacy in mind, and who is accountable for it. It then publishes what it learns so other organisations can reuse those lessons, turning early scrutiny into shared knowledge. Privacy-enhancing technologies, which are tools that let data be analysed or shared while limiting how much personal information is exposed, offer extra protection. But central government support for, and use of, these tools is still limited across the OECD (OECD, 2026[1]).
Making AI accountability enforceable and transparent. All OECD countries plus 12 additional countries and the European Union adhere to the OECD AI Principles (OECD, 2024[12]), and a growing number have developed national principles and advisory bodies for trustworthy AI. These have not yet been translated consistently into enforceable controls. Countries moving toward enforcement have focused on requiring named official responsibility for significant AI systems, advance risk and impact assessments before deploying a system, and post-deployment audits. Illustrating this first point on transparency, the UK’s Algorithmic Transparency Recording Standard (ATRS), which is mandatory for central government departments since 2024, requires a standardised two-tier record published to a central repository, with a plain-language summary backed by a fuller account of the tool’s purpose, data, human oversight and identified risks. Canada’s Algorithmic Impact Assessment (AIA) illustrates a mature example of AI impact assessment: a mandatory questionnaire needs to be completed before a system goes live, which scores both risks and safeguards, and places the system in one of four impact levels with proportionate accompanying required measures. Colombia illustrates an intermediate stage in this transition: its personal data protection regime, the CONPES 4144 (2025[8]) National AI Policy and the Ethical Guide for AI Implementation in Public Entities together establish transparency, risk-assessment and accountability principles for AI systems, alongside emerging sectoral regulation and case law. Colombia does not yet operate a cross-government algorithmic registry and disclosure standard comparable to the UK's ATRS, but the trajectory points toward increasingly binding governance, impact-assessment and transparency requirements for automated systems.
Governing agentic AI as authorised public action under retained accountability. Agentic AI raises the governance stakes mentioned above, because the question shifts from what a system says to what it does. When a system can retrieve data, trigger workflow steps, update records, initiate transactions or co-ordinate action across agencies, it is carrying out action within a public mandate. Countries have responded by requiring that each consequential action have a clear legal basis, a defined permission structure and a named human or institutional sponsor who remains accountable; that actions affecting a person’s rights include a preview, approval or escalation step before they take effect; and that agent actions be logged, open to challenge and capable of being paused or reversed where technically and legally feasible. Controls should be proportionate: a read-only agent that drafts a case summary does not require the same safeguards as one that changes an official record or initiates an eligibility-related transaction. Canada’s Guide on the Use of Agentic Artificial Intelligence (2026) and Singapore’s Model AI Governance Framework for Agentic AI (Version 1.5, 2026) both steer public bodies toward tightly scoped tasks, limited permissions and accountable officials who authorise consequential actions.
Keeping governance adaptive and co-ordinated across borders. No governance framework designed today will be adequate by the time agentic AI is deployed at scale: the technology evolves quickly, risks compound, and their effects routinely cross institutional and national boundaries. Countries have begun responding through more adaptive regulatory design – using sandboxes to test frameworks before hardening them, shifting from prescriptive to outcomes-based oversight, and building systematic horizon scanning into standard governance processes. The OECD Recommendation on Agile Regulatory Governance (2021[13]) sets out the enabling conditions for this kind of adaptability. Spain's AI regulatory sandbox – the first in the EU, developed with the European Commission and open to other Member States – shows how one national pilot can begin to shape a common regional supervisory approach and how adaptability and co-ordination can be designed together rather than pursued in parallel. Still, cross-border co-ordination remains limited: only 13 of 36 OECD countries engage cross-border partners on AI in government beyond participation in international fora (OECD, 2026[1]). Finally, public legitimacy underpins all of this: accountability frameworks that citizens can see, understand and contest are what build the trust that makes data- and AI-enabled government politically sustainable at scale. Belgium's 2024 citizens' panel on AI and France's Agora platform illustrate how deliberation on AI governance can be broadened beyond expert circles, building the informed public consent that enables, rather than progressively constrains, more capable government.
Forward looking issues
All OECD countries have committed to trustworthy AI through adherence to the OECD AI Principles, but the mechanisms that convert those commitments into practice – transparency standards, advance risk assessments, post-deployment audits and routes for citizens to challenge AI-assisted decisions – remain the exception. Two areas emerge as priorities, both for national systems and for the OECD's future work in supporting them:
Understanding and measuring trustworthy AI in practice. The most important near-term step is a comparative baseline of how countries are implementing the OECD Framework for Trustworthy AI in Government: measuring whether its enablers, guardrails and engagement mechanisms operate in day-to-day practice rather than only on paper. The purpose is not additional reporting but a diagnostic tool that directs effort toward making existing instruments work where they exist and adopting them where they do not. That baseline should extend to citizen involvement in live services, not only to strategy consultations and set-piece deliberative exercises, treating feedback and complaint mechanisms as a routine governance feature rather than an exceptional one.
Sharing lessons on successful adaptive and cross-border governance. Agentic AI and cross-border data use are spreading faster than governance can adapt, so the practical question is how countries can align their oversight before national rules settle into approaches that cannot keep up with the technology or work across borders. The OECD is well placed to develop comparative evidence on what outcomes-based, adaptive oversight looks like in practice, for example how sandbox findings are recognised across borders, how supervisory judgements are shared, and which regulatory designs have proved able to accommodate technologies that change after deployment. This is where exchanging and convening across countries matters most, by supporting the practical convergence of supervisory approaches that national action alone cannot achieve.
3.2.1. Introduction
The objectives set out in Chapter 2 require public servants to work across organisational boundaries, collaborate with citizens and external partners, harness AI and digital technologies, exercise judgement under uncertainty and continuously improve policies and services. Yet the capabilities governments increasingly require are not always those that civil service systems were originally designed to develop. Recognising this, civil service reform has become a central component of public administration reform agendas. The Government Unstuck Survey shows that among countries that have adopted or are developing public administration reform strategies, the large majority (25 out of 32 responding countries) include reforms to civil service frameworks and leadership capability as core priorities, (OECD, 2026[14]), and while efficiency and productivity remain a key objective of these reforms, the results suggest that productivity depends on deploying people differently rather than simply employing fewer, as relatively few continue to rely on across-the-board staffing reductions (12 out of 40 countries responding to the survey). The survey also shows that resistance to organisational change remains one of the most commonly reported barriers to reform, and only around half of responding countries use performance management systems to incentivise the achievement of efficiency objectives, while even fewer designate senior leaders to champion reform across government. (OECD, 2026[14]) These findings suggest that the challenge is increasingly one of leadership, incentives and culture, and reflect growing recognition that a new model of civil service is fundamental to successful government transformation.
Evidence from the 2024 OECD/EU Survey of Public Servants reinforces this picture (OECD, 2025[15]). Across nearly 57 000 public servants, employees report high levels of commitment to public service and willingness to contribute beyond formal job requirements, yet they are considerably less positive about the organisational conditions needed for transformation. While perceptions of overall organisational performance are only moderate, employees are least positive about their organisations' ability to respond quickly when change is needed and to use resources efficiently (see Figure 3.2). Only around four in ten believe senior leaders lead organisational change effectively or act on employee feedback (OECD, 2025[15]). Around half report that organisational rules and procedures make it harder to work effectively, and 52% believe that underperforming employees typically remain in post without improving (OECD, 2025[15]). At the same time, innovation is one of the weakest dimensions of organisational culture, despite being the single strongest driver of perceived organisational performance (OECD, 2025[15]). Together, these findings point to a systemic problem: While public servants are generally committed to their work, they feel the leadership, cultures and workforce systems within which they operate are not yet consistently aligned with the behaviours required for innovation and transformation.
Note: The figure presents the EU8 average for the underlying items of the organisational performance index. EU8 includes all project countries. Positive responses refer to the combined share of respondents selecting ‘strongly agree’ or ‘agree,’ neutral responses those selecting ‘neither agree nor disagree,’ and negative responses those selecting ‘strongly disagree’ or ‘disagree.’ Averages are normalised, giving equal weight to each country regardless of sample size.
* The original item question, formulated as “has rules and procedures in place that make it difficult and complicated to work effectively and efficiently,” has been reverse-coded and reworded in the visualisation to align with the direction of other items.
Source: OECD (2025[15]), Workforce Insights from Central Governments: Findings of the 2024 OECD/EU Survey of Public Servants, OECD Publishing, Paris, https://doi.org/10.1787/2f9080b1-en.
To address this challenge, many governments are taking active measures to change how they lead, organise and develop the public workforce. This section therefore focuses on four interconnected priority areas of practice. First, it looks at how governments align leadership, accountability and incentives with transformation, innovation, long-term stewardship and the delivery of outcomes. Second, the section focuses on the civil service cultures that encourage experimentation, responsiveness to citizens, continuous learning and integrity, creating environments where responsible innovation is expected and supported. Third, how governments are building the workforce of the future, developing the professions, capabilities and career models required for an increasingly digital, AI-enabled and people-oriented public service. Finally, how governments are modernising the HR function, equipping it to act as a steward of government capability through better workforce planning, talent management and workforce intelligence. This section tackles these four key areas in that that order and highlights defining practices for each (see Table 3.3).
Defining practices for Skilled and empowered public servants
3.2.2. Accountability and incentives to drive the change
Key issues
Accountability in most governments remains largely organised around individual organisations, vertical reporting lines and the delivery of discrete outputs. While these arrangements may support administrative control, they are often poorly suited to the cross-cutting reforms required to simplify government, improve responsiveness and deliver better outcomes for citizens. Furthermore, authority, resources and responsibility are frequently distributed across multiple organisations, making it difficult to establish who is accountable for delivering system-wide change. These structural challenges are reinforced by career systems that often incentivise behaviours different from those governments increasingly demand. Senior public servants are typically selected and promoted because of their ability to navigate (rather than simplify) complex administrative systems, manage risks and ensure compliance. While these capabilities remain essential, they can also reinforce behaviours that favour stability over transformation and encourage risk aversion.
The result is that many governments increasingly ask leaders to deliver transformational change while accountability and incentive systems primarily ensure procedural compliance and organisational control. Every element of the direction of change is delivered by people working inside these systems, which is what makes the mismatch consequential. Governments are therefore implementing new measures that redefine what senior leaders are accountable for, and how accountability is created. Rather than relying solely on hierarchical reporting and individual organisational performance, the practices highlighted here seek to strengthen collective ownership of reform, distribute and clarify responsibility for cross-government priorities, increase visibility of delivery, and better align leadership incentives with desired behaviours such as collaboration, innovation and long-term stewardship. The practices below illustrate a growing set of governance innovations that seek to move accountability beyond process compliance towards enabling and sustaining public sector transformation.
Defining practices
Creating collective accountability for transformation. Delivering complex reforms requires accountability arrangements that extend beyond individual organisations. Governments are increasingly creating governance structures that give senior leaders shared responsibility for cross-government priorities, encouraging collaboration, peer challenge and collective problem-solving. Ireland's Civil Service Management Board (CSMB), established to oversee implementation of the Civil Service Renewal Plan, illustrates this approach by assigning each Secretary General responsibility for leading one of the Plan's reform priorities on behalf of the entire civil service, irrespective of departmental boundaries (OECD, 2023[16]; OECD, 2026[14]). This transformed the Board from a co-ordination forum into a collective executive body for reform, creating shared ownership and stronger incentives for collaboration. The model has since evolved into the broader Public Service Leadership Board, extending these principles across the wider public service. Iceland's Efficiency in State Operations reform demonstrates a complementary approach by involving more than 90 agency heads in identifying efficiency measures within their own organisations, creating ownership from the outset. Accountability for implementation is then reinforced through a cross-ministerial taskforce led by the Prime Minister's Office and Ministry of Finance, with progress reported publicly and integrated into the budget process, ensuring sustained collective responsibility for delivery.
Aligning individual leadership incentives with transformation objectives. Governments are increasingly embedding responsibility for transformation into the formal expectations placed on senior leaders by aligning mandates, authority, performance management and career systems with broader public outcomes. Canada does this through Prime Ministerial mandate letters that make reform priorities an explicit leadership responsibility. Korea's Proactive Administration regulations complement this by giving public servants greater authority and confidence to act in situations where legal interpretation is uncertain, linking responsibility with the ability to make decisions. The United Kingdom's 2025 Senior Civil Service Performance Management Framework strengthens accountability by assessing senior leaders against organisational performance, efficiency and the delivery of ministerial priorities, while France's creation of the Délégation interministérielle à l'encadrement supérieur de l'État (DIESE) modernises senior leadership management by promoting merit-based appointments, mobility and stewardship across government (UK Civil Service, 2026[17]). It represents a shift from a closed, career-based model toward a more dynamic leadership system focused on capability, performance and cross-government contribution, thereby reinforcing the idea that senior leaders are not just administrators of their own organisations, but stewards of the broader public service. Together, these approaches reinforce the expectation that senior leaders are responsible not only for managing their own organisations, but for driving broader public sector transformation.
Strengthening accountability through transparency and public recognition. Governments are also reinforcing accountability by making reform performance more visible to citizens and using transparency as a driver of continuous improvement. New Zealand's Government Targets publicly identify both the Lead Minister and Lead Chief Executive responsible for each cross-government priority, with quarterly reporting creating regular opportunities for political scrutiny and public accountability (NZ DPMC, 2026[18]). The United Arab Emirates' Zero Government Bureaucracy Programme complements this approach through public benchmarking and annual awards that recognise ministries achieving the greatest progress in simplifying procedures and improving services (UAE, 2024[19]). By combining transparency with reputational incentives, these approaches encourage sustained leadership attention, healthy competition and continuous improvement in delivering reform.
Forward looking issues
The defining practices above demonstrate important progress in aligning leadership accountability to the transformational objectives inherent in the Government Unstuck Agenda. Looking forward, governments will have to regularly review their policies and practices in this area to ensure the right balance between autonomy and compliance in a fast-changing world, where the lines between public and private, and human and machine, are increasingly blurred. The following are a few issues that may benefit from further work by government and the OECD.
Broadening accountability to citizens. Most accountability systems continue to focus on upward accountability within government, with senior leaders reporting to ministers, central agencies and legislatures. As governments seek to rebuild trust and strengthen democratic legitimacy, accountability will also need to become more outward facing. This could involve greater transparency on reform progress, more meaningful reporting of outcomes that matter to citizens, stronger mechanisms for citizen feedback, and new approaches to involving citizens in monitoring and evaluating reforms. Future accountability systems may need to treat citizens not only as beneficiaries of reform but as active participants in holding governments to account for its delivery and impact.
Balancing political and administrative accountability. Many of the most pressing public sector transformations, from fiscal sustainability to digital transformation and climate adaptation, require sustained implementation over the medium and longer term. Yet accountability mechanisms are often closely tied to electoral and budget cycles, creating incentives to prioritise short-term delivery over long-term reform. Governments may therefore need to explore new institutional arrangements that provide continuity of leadership, maintain political commitment across administrations, and preserve accountability for long-term objectives while respecting democratic transitions. This may also require a focus on effective working relationship between ministers and the senior civil service. Future reforms may need to strengthen ways of working, knowledge and skills on both sides, so that political leaders understand how to steer the administration and senior public servants can advise, implement and ensure system stewardship. Investing in this relationship is likely to become increasingly important for both responsiveness and government performance. The OECD is examining these questions and exploring how this critical interface can be strengthened to support both responsiveness and long-term stewardship. Related work is also looking at emerging programmes for the training of ministers and other political leaders, including through administrative learning institutions, utilising the OECD’s Network of Schools of Government (NSG).
Preparing accountability systems for the age of AI. Artificial intelligence, and particularly agentic AI, has the potential to fundamentally reshape accountability in government. AI could enable real-time monitoring of implementation, early identification of delivery risks, and more dynamic performance management, reducing reliance on retrospective reporting. At the same time, as AI agents begin to support or automate decisions, governments will need to redefine accountability for those decisions, ensuring that responsibility remains clear, transparent and attributable. Future accountability frameworks will need to harness AI's potential to strengthen implementation while preserving human judgement, oversight and public trust. The OECD’s work on trustworthy and accountable AI systems, described in section above, will continue to explore this emerging issue, with a view to not only ensuring the right structural safeguards and frameworks are in place, but also that civil service leadership is supported with the right knowledge, incentives and responsibilities to ensure accountability for both progress and safety in a world where human and digital agents are increasingly working together.
3.2.3. Public service culture that encourages innovation, citizen responsiveness and integrity
Key issues
Many public administrations continue to be shaped by working cultures developed for a context where consistency, procedural fairness and risk control were rightly prioritised to ensure impartiality, legality and accountability. These values remain fundamental. However, when organisational incentives place greater weight on avoiding mistakes than improving outcomes, public servants may reasonably conclude that following established processes is safer than testing new approaches, collaborating across organisational boundaries or adapting services to citizens' needs. As a result, innovation becomes exceptional rather than expected, learning gives way to compliance, and responsiveness is constrained by institutional routines rather than driven by public value. Cultures that reward navigating complexity rather than removing it are among the reasons government stays complicated. Recent OECD survey evidence suggests that innovation remains one of the weakest aspects of organisational culture across central governments (OECD, 2025[15]). Fewer than half of employees hold positive views of their organisation's innovation climate, making it one of the lowest-scoring organisational dimensions in the survey (see Figure 3.3). Employees are particularly sceptical that their organisations learn from past successes and failures, encourage them to find new ways of working, or recognise and support new ideas. Only 31% feel their organisation has a sufficiently flexible regulatory framework to incorporate new ideas (OECD, 2025[15]). At the same time, innovation climate emerged as the single strongest driver of organisational performance across all factors examined. (OECD, 2025[15]).
Note: The figure presents the EU8 average for the underlying items of the innovation climate index. EU8 includes all project countries, Positive responses refer to the combined share of respondents selecting ‘strongly agree’ or ‘agree,’ neutral responses those selecting ‘neither agree nor disagree,’ and negative responses those selecting ‘strongly disagree’ or ‘disagree’. Averages are normalised, giving equal weight to each country regardless of sample size.
Source: OECD (2025[15]), Workforce Insights from Central Governments: Findings of the 2024 OECD/EU Survey of Public Servants, OECD Publishing, Paris, https://doi.org/10.1787/2f9080b1-en.
Addressing this requires deliberate leadership: shaping incentives, role-modelling behaviours, creating psychological safety for experimentation, involving citizens more directly in improving public services, and embedding integrity as a shared professional value that enables better decisions rather than simply enforcing compliance. Increasingly, governments are recognising that effective integrity systems help public servants exercise sound judgement, manage risks proportionately and innovate responsibly, rather than minimise all forms of risk through compliance. Recent OECD work points to a broader shift from rules-based, compliance-focused approaches towards more risk-based and results-oriented integrity systems, recognising that integrity is one of the essential enablers of responsible innovation and sustained public trust (OECD, 2026[20]).
The practices highlighted below show how government are taking steps to create protected spaces for experimentation, support the scaling of effective pilot projects and signal that responsible experimentation and innovation is expected. They show how some governments are experimenting with different ways of working, including embedding support for human centred design and behavioural science within government operations. And they identify new approaches to integrity, to ensure that civil servants have the support to take decisions and drive change responsibly.
Defining practices
Creating the conditions for experimentation and responsible risk-taking. Some countries establish clear frameworks that allow public servants to test new approaches while managing risks proportionately. Several examples show how countries are already supporting these conditions in practice. The UK’s Test, Learn and Grow initiative is a GBP 100m programme to build capacity to design, test and scale responses to complex policy and delivery challenges via partnerships to support place-based working, an agile “test and learn” policy framework”, and multi-disciplinary teams working in partnership with local authorities across England. Finland’s experimentation lab for education enables real‑world testing of policies before wider rollout, while in Belgium strategic foresight simulations provide safe, anticipatory spaces to test policy choices (Finland Ministry of Education and Culture, 2026[21]; Government of Flanders, 2024[22]). In other examples, France’s ‘learn by doing’ public transformation campus embeds experimentation in practice, giving public servants a structured environment to iterate in real time, Denmark’s regulatory sandbox for AI allows authorities to test innovative solutions while managing legal and ethical risks, while the UK’s National Digital Twin Programme enables policy and infrastructure decisions to be tested in virtual environments before implementation (DITP, 2024[23]; Danish Agency for Digital Governance, 2025[24]; GOV UK, 2023[25]) . Colombia is implementing a territorially differentiated Public Sector Experimentation Strategy which integrates various design approaches as a methodological framework for national and subnational entities to design, test and adjust interventions under real-world conditions. At a cross-border level, the OECD Gov2Gov Innovation Incubator extends this model by providing a shared space for governments to experiment collaboratively, adapt solutions, and transfer proven approaches across contexts while managing risk through peer exchange and prototyping (OPSI, 2025[26]).
Embedding citizen-centred ways of working across government. Many OECD countries incorporate approaches that bring public servants closer to citizens' experiences and needs. This includes promoting human-centred design, behavioural insights, agile methods, co-design, deliberative approaches and tools such as service audits that help identify where citizens get stuck in their interactions with government. OECD Countries are currently working towards embedding design to support the delivery of human-centred public services. Ireland’s Action Plan for Designing Better Public Services aims to strengthen simplification through embedding design mindsets, co-design, prototyping and user journey mapping across the public service. Finland’s Social Security Institute (Kela) developed a tailored service design approach for young people, to design public services around young people’s specific home-support needs. The Netherlands has institutionalised behavioural science by embedding behavioural teams across ministries, supporting them through a cross-government network, and integrating behavioural analysis into the government's core policy development methodology. Meanwhile, Australia has embedded behavioural science at the centre of government through the Behavioural Economics Team of the Australian Government (BETA), combining rigorous research on high-priority policy issues with a systematic programme to build behavioural capability across the Australian Public Service (OECD, 2024[27]).
Incorporating a risk-based, results-oriented approach to integrity frameworks. In some countries, public integrity systems are evolving from compliance-heavy approaches that treat all activities and public servants alike towards risk-based systems that focus oversight where integrity risks are greatest. Increasingly, governments are using AI, machine learning and advanced analytics to identify fraud risks, target audits and detect anomalous behaviour, allowing oversight resources to be deployed more effectively. Countries such as Brazil and Spain are using AI and advanced analytics to strengthen fraud detection and audit, while Finland has digitised conflict-of-interest declarations to improve verification and reduce administrative burdens. Such approaches not only improve the effectiveness of integrity systems but also contribute to a healthier organisational culture by reducing unnecessary controls, enabling greater professional judgement, and signalling trust in public servants to innovate responsibly within clear ethical boundaries. Leaders are therefore encouraged to review existing controls, reporting requirements and oversight mechanisms to identify the hidden costs of compliance, ensuring that integrity frameworks remain proportionate, support innovation, and strengthen ethical behaviour rather than encouraging defensive administration. At the same time, building a strong culture of integrity requires continued investment in merit-based human resource management, ethical leadership, open organisational cultures and capability building (OECD, 2026[20]).
Forward looking issues
The defining practices above define how some governments are beginning to reshape public service culture for a more adaptive and innovative public sector. Yet risk aversion and citizen distance continue to define significant aspects of public service cultures. This will be even more complicated with rapid technological change and rising expectations of government. The following are a few issues that may benefit from further work by government and the OECD.
Creating cultures that manage risk rather than avoid it. Risk aversion remains one of the greatest barriers to public sector innovation. While integrity, accountability and stewardship require appropriate controls, future public administrations will increasingly need to distinguish between unacceptable risks and responsible experimentation. This implies moving beyond cultures that reward procedural compliance towards environments where calculated risk-taking, learning from failure and continuous adaptation are recognised as essential components of effective public service. Developing clearer organisational risk appetites and equipping leaders to manage innovation responsibly will become increasingly important. OECD work on leadership, public sector innovation, integrity and transformation will continue to explore different elements of this challenge. Future editions of the OECD/EU Survey of Public Servants could also be focused on assessing this issue in more detail, providing new insights and comparative evidence that countries can use to identify root causes and solutions.
Embedding continuous citizen feedback into organisational culture: Future civil service cultures will need to become more outward-looking, treating citizens not only as service users but as continuous partners in learning and improvement. Digital technologies, behavioural insights, social listening and participatory approaches create opportunities for public organisations to receive real-time feedback on how policies and services are experienced in practice. The challenge is no longer simply consulting citizens periodically, but embedding citizen perspectives into everyday decision making, performance improvement and organisational learning. Civil services that continuously adapt based on evidence from citizens are likely to become both more innovative and more trusted. While many OECD countries are implementing new approaches and technology in these areas, the OECD also has various tools that can support countries to identify these issues, across multiple work streams, including in open government, public sector innovation, service design, and regulatory governance, among others. Future work could bring these streams together into a common framework that could enable benchmarking and comparison with outcome indicators such as the OECD trust survey and the OECD/EU survey of public servants, to create the multidimensional perspective needed to assess this topic.
Sharpening the focus on performance: It is hard to incentivise innovation without a more direct and explicit link to performance management systems. Effective performance management can help governments translate strategic priorities into clear expectations, measurable targets and shape expectations for a working culture focused on performance and delivery. This is particularly true in an era of fiscal constraint and heightened public scrutiny of the civil service. Robust performance systems can help demonstrate that public resources are being used effectively and that there are systems in place to encourage and support the types of proactive and innovative behaviour that public service users should expect. Future work could examine how to link performance management systems more directly and explicitly to transformation outcomes. This could be done as part of broader work on identifying and strengthening the levers available to senior leaders to carry out transformation.
3.2.4. The workforce of the future
Key issues
The capabilities governments require are changing faster than the workforce systems used to recruit, develop and organise public servants. Artificial intelligence, digital technologies and new expectations around citizen engagement are reshaping how governments operate, requiring governments to redefine not only the skills of individual employees, but the structure of the professions that make up the public service. Many governments have responded by investing in learning and development. While this remains essential, training alone will not be sufficient. The future civil service will require structural changes – redesigning the occupations and professions on which governments increasingly depend. This means identifying the capabilities that will become strategically important, creating recognised professional communities with common standards and career pathways, and supporting workforce transitions from declining to emerging roles.
Despite growing recognition of these challenges, many governments have yet to develop systematic approaches. In 2022, 43% of responding OECD countries reported having no dedicated reskilling strategy, while 57% reported providing no structured reskilling training, instead expecting employees to acquire new capabilities informally through their day-to-day work (OECD, 2023[28]). National Schools of Government have a critical role to play in preparing the future public service, but recent OECD evidence suggests that many are not yet equipped for this transition. While almost all responding schools that provide AI training cover AI literacy (96%) and AI ethics (91%), far fewer offer learning on how AI is reshaping professional roles, organisational change, leadership or specialist AI capabilities. Schools rate their own readiness to teach and use AI at only 5.2 out of 10, citing the pace of technological change, limited budgets and shortages of AI expertise as major constraints (OECD, 2026[29]). These findings suggest that governments now need to move beyond teaching public servants how to use AI towards preparing them to work differently because of AI, redesigning professional roles so that technology complements rather than replaces human judgement, creativity and collaboration. Being prepared in a digital world is, in the end, as much a question about people than it is technology (see Section 3.1 on “Trusted and connected digital technologies and data” for a discussion of the technical practices countries have put in place to increase their AI capabilities).
The defining practices below show how governments are beginning to redefine the future civil service by establishing common capabilities across the workforce, building professional communities around emerging occupations, and creating more dynamic approaches to workforce planning and career development.
Defining practices
Redefining the foundational capabilities expected of all public servants: Governments are increasingly establishing common capability frameworks that define the knowledge, skills and behaviours expected across the entire civil service, embedding them into recruitment, learning, career development and performance management. These frameworks signal which capabilities are becoming universal expectations, while creating a common language for workforce planning and talent management. Latvia’s recent focus on AI skills development illustrates how governments are beginning to redefine foundational capabilities in practice. Through initiatives such as public administration training programmes and national AI guidelines, the government is promoting competencies including critical use of AI tools, prompt design, data assessment, and ethical application. (Republic of Latvia, 2026[30]) Similarly, the United Kingdom has developed an AI Skills Competency Framework that defines differentiated AI capabilities for technical, responsible and ethical, and non-technical domains, as well as for managers, mid-level and entry-level professions (Government of the United Kingdom, 2025[31]) (see Box 3.3). Together, these approaches illustrate how governments are embedding digital and AI capabilities as core elements of modern public service professionalism rather than treating them as specialist technical skills.
The Framework provides a structured approach to defining the capabilities needed to use, oversee and integrate artificial intelligence across the workforce. Developed as part of the UK Government’s AI Skills Tools Package, the framework organises AI competencies into three domains: technical skills, responsible and ethical skills, and non-technical skills, and maps them across three levels of responsibility: entry-level, mid-level and managerial roles. Rather than focusing only on AI specialists, the framework is designed for the broader workforce and emphasises practical competencies such as AI literacy, prompt writing, evaluation of AI outputs, workflow automation, ethical oversight and strategic governance.
The UK framework illustrates a comprehensive effort to embed AI in everyday service delivery and decision making by providing common language for recruitment, workforce planning, learning and career development across the whole civil service. The framework sits alongside two complementary tools within the UK’s AI Skills Tools Package: an AI Skills Adoption Pathway Model, which links skill needs to different stages of organisational AI adoption, and an Employer AI Adoption Checklist, which helps organisations assess readiness, identify capability gaps and plan implementation. Together, these tools create an integrated approach that connects individual skills, organisational capability and workforce strategy.
Source: UK Government (2025[31]), AI Skills Competency Framework, https://www.gov.uk/government/publications/ai-skills-for-the-uk-workforce/ai-skills-tools-package#the-ai-skills-framework-mapping-ai-skills-by-job-level.
Identifying and prioritising the professions governments will increasingly depend on. Governments are increasingly adopting profession-based workforce models to identify, co-ordinate and invest in the specialist capabilities most critical for future government performance. Rather than treating emerging skills as isolated recruitment challenges, these approaches recognise that capabilities such as AI, data, digital, commercial expertise and service design require recognised professional identities, common standards and structured career pathways. The United Kingdom's Civil Service Professions model illustrates this approach by organising the workforce into more than thirty cross-government professions, each with common competency standards, professional leadership and career development frameworks. More recently, the UK's AI Skills Competency Framework has extended this model by identifying a new generation of AI-enabled roles, including AI leaders, business experts, AI specialists and system users, recognising that successful adoption depends on building professional capability across government rather than simply recruiting technical experts. Singapore has taken this approach a step further through its Competency-Driven Growth initiative, mapping every public sector role against common competency frameworks that underpin personalised learning, career development and workforce mobility through the One Talent Gateway platform (Government of Singapore, 2024[32]). Together, these approaches demonstrate how governments are moving beyond isolated training initiatives towards building sustainable professional communities capable of evolving as technologies, citizen expectations and public sector priorities continue to change.
Accelerating workforce transitions from declining to emerging capabilities: As AI and digital transformation reshape public sector work, governments will increasingly need to support workforce transitions rather than relying on natural attrition or external recruitment alone. Australia has developed one of the most advanced approaches through the Australian Public Service Commission's evidence-based reskilling framework (OECD, 2023[28]). Using workforce analytics, job family frameworks and labour market data, it identifies occupations likely to decline, maps transferable skills to growing occupations, and designs targeted reskilling pathways for affected employees. By identifying viable career transitions before workforce disruption occurs, the approach demonstrates how strategic workforce planning can help governments retain valuable institutional knowledge while preparing the civil service for changing capability needs.
Forward looking issues
The defining practices above demonstrate how governments are beginning to redesign the future public service workforce. Yet the pace of technological change is raising fundamental questions about the future of work in government that few countries have yet resolved. The following issues highlight some of the key frontiers for future government practice and OECD support.
Using AI to augment public servants’ capabilities. AI is already reshaping the work of many public servants, but the direction of that change varies by country and context (see Section 3.1 on “Trusted and connected digital technologies and data”). When managed well, AI can augment human capability, giving public servants powerful tools to improve analysis, problem-solving, citizen engagement and complex decision making. It can also be used to standardise decision making and increase managerial control, narrowing the scope for professional judgement, and in some cases, it may eliminate certain roles, which needs to be carefully managed as part of an overall workforce strategy. Much current investment focuses on teaching public servants to use AI, but as routine tasks are automated, the "real" AI skills may turn out to be distinctly human capabilities such as judgement, ethical reasoning, communication and management, and citizen engagement. Governments will need to anticipate how tasks are redistributed, occupations transformed and new professions created, while ensuring public servants retain the expertise to question, interpret and improve AI-generated outputs rather than becoming passive users of them. Senior leaders in particular will face decisions on the workforce and on policy, from redesigning roles to determining where human oversight is essential. The OECD is developing analysis on AI skills in the public service, covering both the skills needed to design and implement AI and the wider workforce implications for leadership and shifting roles; including a survey of schools of government on AI training and institutional readiness; and support to countries in managing AI-driven workforce change, including downsizing and equipping leaders to direct the transition.
Developing leaders for the future. Improving the quality of senior leadership is a continuous commitment, requiring governments to ensure the best people reach the most consequential roles and are developed on their way there. This means working in closer partnership with leadership programmes and schools of government, whose curricula must evolve as quickly as the challenges leaders face; these institutions must re-think what learning and teaching are for leadership. Most leadership capability is built on the job, and formal programmes matter less than the quality of the experiences leaders accumulate. Traditional classroom-based approaches are insufficient: leaders need hands-on experience through action learning, stretch assignments, rotations and simulations grounded in the real problems of governing. All of these can be integrated into strategies of schools, who must rethink their functioning. Governments that treat leadership development as a strategic investment rather than a training obligation will be better placed to sustain transformation over time. The OECD Network of Schools of Government (NSG) provides a platform to drive this shift, enabling administrations to design and pilot new approaches and redefine their function together, while enabling a more rapid scale up of this newer form of learning through sharing and co-operation on the approach.
Dynamising public service professions. The pace of technological change is raising questions about whether traditional occupational and career structures remain fit for purpose. Governments will continue to require recognised professions with clear standards, career pathways and professional identities, but these systems will need significant flexibility to accommodate rapidly evolving skills and changing ways of working. The challenge will be to find a balance between creating new professions and ensuring existing occupational frameworks can adapt more quickly by incorporating new competencies, supporting movement between related professions, and enabling public servants to refresh their expertise throughout their careers. This will require governments to balance the stability and coherence provided by professional structures with the agility needed to respond to changing technologies, citizen expectations and labour market demands.
Expanding the range of employment models for more flexibility. The traditional model of a lifelong career civil service may become less common, particularly in technical and fast-moving fields where expertise evolves rapidly and governments compete with other sectors for talent. At the same time, relying solely on external consultants risks weakening institutional capability and increasing dependency on outside expertise. The future challenge will therefore be to develop a broader range of employment models that sit between these two extremes. More flexible contractual arrangements, fixed-term appointments, secondments, fellowships and rotational exchanges could enable governments to attract specialist expertise for defined periods while preserving a strong core of career public servants who provide continuity, institutional memory and stewardship. At the same time, a growing number of OECD countries face sustained pressure to restore public finances. Because workforce costs account for a significant share of public expenditure, many fiscal consolidation efforts will inevitably involve the civil service, and workforce decisions are among the most direct routes to the restored margins of manoeuvre the direction of change describes, as well as among the easiest to get wrong under pressure. The OECD's work on Restoring Public Finances examines the approaches countries have taken to workforce reduction in recent years, drawing early conclusions on which will likely be sustainable and which could cost governments capability they later had to rebuild (see Box 3.7). Governments already possess a range of workforce management levers, including recruitment controls, natural attrition, redeployment, organisational restructuring and greater use of technology. Building the workforce of the future will likely require a more coherent and strategic approach to workforce resizing so that these types of decisions are grounded in plans rather than taken too hurriedly in crisis points. Future OECD work in this area could include developing a practical methodology to support workforce resizing and restructuring decisions, helping governments identify where capacity can be reduced, redeployed or strengthened while safeguarding the core capabilities needed to deliver on strategic objectives and sustain transformation over the longer term.
3.2.5. Human resource management as a strategic enabler of government transformation
Key Issues
Many governments are seeking to build more agile, skilled and future-ready workforces, yet the human resource management (HRM) functions responsible for enabling this transformation often remain focused on administrative compliance, transactional processes and position management. In many governments, HRM systems were designed for a more stable environment, where workforce needs changed gradually, career paths were predictable and workforce management was primarily concerned with maintaining established staffing structures. Today, however, governments need HRM functions capable of anticipating future capability needs, attracting scarce talent, enabling workforce mobility, supporting career transitions and providing leaders with the data and insight needed to make strategic workforce decisions. HRM is therefore evolving from an administrative support function into a strategic enabler of government transformation and a steward of government capability.
Recent OECD evidence suggests that governments broadly recognise this increasing urgency of this challenge but have yet to update and align many of the systems needed to respond effectively. While 70% of OECD countries identify improving HR data and analytics as a priority, only 24% are explicitly seeking to strengthen the strategic capability of the HR function itself through the creation of new HR roles (OECD, forthcoming[33]). At the same time, governments continue to face persistent challenges in attracting specialist talent, enabling workforce mobility and supporting large-scale workforce transitions, despite recognising these as increasingly important for public sector performance.
The practices below illustrate how governments are beginning to modernise HR by strengthening talent attraction and recruitment, creating more agile internal labour markets, professionalising the HR function itself, and investing in workforce intelligence and analytics. Together, these approaches reflect a broader shift from HRM as an administrative service towards HRM as a strategic partner in building the workforce governments will need for the future.
Defining practices
Attracting critical talent. Governments are increasingly redesigning recruitment systems to compete more effectively for scarce talent while maintaining merit-based principles. This includes simplifying recruitment processes, strengthening employer value propositions, broadening recruitment channels and updating employment frameworks to attract emerging professions. In Canada, reducing hiring times is a key priority of the Public Service Commission, who aims to reduce average hiring time by 22% by simplifying recruitment processes and improving flexibility for managers (Government of Canada, 2025[34]). Greece has used behavioural experimentation to test how different public service messages influence recruitment, finding that messages centred on civic duty were 47% more effective than generic recruitment campaigns. France has modernised recruitment for digital professions by updating occupational classifications and remuneration frameworks, creating clearer career structures and more competitive salary arrangements for specialist talent (OECD, 2026[35]).
Enabling agile internal talent markets. Rather than treating employees as permanently attached to organisational units, governments are increasingly creating mechanisms that allow talent to move more easily across government in response to changing priorities. Australia has developed one of the most comprehensive approaches through its APS Mobility Framework, including Temporary Opportunities and APS Surge Reserve on the APSjobs platform and oneAPS Opportunities platform, enabling temporary deployments, project-based assignments and rapid cross-government mobilisation during emergencies while strengthening workforce capability and collaboration. Together these initiatives demonstrate how HR can create a more dynamic internal labour market that improves both organisational agility and employee development (Australian Public Service Commission, 2021[36]).
Professionalising the HR workforce. As HR assumes a more strategic role, governments are investing in new capabilities for HR professionals, including workforce planning, organisational development, workforce analytics, strategic talent management and change leadership. Ireland's Civil Service People Standards Framework defines these capabilities explicitly, creating professional expectations for modern HR practice. Similarly, the UK Government People Group has made building a high-performing HR profession a central pillar of its Civil Service People Plan, strengthening HR business partnering, analytical capability and strategic workforce support across government (Government of Ireland, 2025[37]; United Kingdom Government People Group, n.d.[38]).
Developing workforce intelligence through data and analytics. Modern HR increasingly depends on timely workforce intelligence rather than administrative reporting. Greece is developing a centralised Human Resources Management System alongside an AI-enabled workforce planning tool that combines workforce supply, organisational structures and workload indicators to anticipate capability gaps, support recruitment decisions and enable evidence-based workforce planning across almost 300 public organisations. These investments demonstrate how integrated HR data and analytics can strengthen governments' ability to anticipate workforce needs and manage transformation proactively rather than reactively.
Forward looking issues
The defining practices above demonstrate how governments are beginning to transform HR from an administrative function into a strategic steward of public sector capability. Yet many of the opportunities created by AI, workforce analytics and strategic workforce planning remain largely unexplored. The following issues highlight some of the key frontiers for future government practice and OECD support.
Moving towards intelligent workforce stewardship. Artificial intelligence has the potential to transform the HR function itself. While many governments are beginning to use AI to automate recruitment, streamline administrative processes and improve workforce analytics, the greatest opportunity lies in using AI to strengthen strategic workforce stewardship. AI could help governments anticipate future skills needs, model the workforce implications of policy choices, identify emerging capability gaps, match people to new roles and personalise learning and career development at scale. Future work in this area could help governments demonstrate concrete productivity gains from AI use in the public workforce and support the reallocation of talent and resources toward higher-value work in line with strategic objectives.
Shaping the future workforce. As governments face fiscal constraints, demographic change and rapid technological disruption, workforce choices are becoming some of the most important strategic decisions they make. Questions about where to invest in capability, which occupations to grow or reduce, how AI should reshape work, and how to balance internal capability with external expertise can no longer be addressed after policy decisions have been taken. Future work in this area could demonstrate how workforce planning can move beyond forecasting staffing needs to become a core tool for delivering transformation agendas. This could include linking workforce decisions directly to government priorities, missions and reform objectives.
Strengthening the capacity of the HR function as an internal transformation partner. Traditional HR functions in many governments remain heavily focused on compliance, processes and risk management, often limiting their ability to respond quickly to emerging organisational challenges. Looking ahead, public administrations may need a more agile and demand-driven HR model, operating less as a central administrative function and more as an internal ‘on-demand’ consultancy. This type of model could deploy multidisciplinary teams to work directly with leaders on critical workforce and organisational challenges, from improving performance and redesigning operating models to addressing skills gaps and strengthening employee engagement. Future work in this area could support governments to transform their HR function to emphasise solving business problems, testing solutions in real-world settings and scaling successful approaches across government. The HR function has long asked for ‘a seat at the table’ – this work would deliver on that ambition by positioning HR as a strategic enabler of transformation, with greater emphasis on organisational effectiveness and delivery outcomes, while reducing the time and resources devoted to low-value compliance activities.
Redesigning the organisation of HR services. Responsibility for HR management is distributed differently across OECD countries, with some functions delivered centrally and others delegated to ministries and agencies. While these arrangements often reflect historical governance choices, future work in this area could assess whether they continue to provide the best balance of efficiency, capability and service quality. One area for future exploration is the development of more specialised HR service models, where centres of expertise provide high-quality workforce, organisational development and associated services across government on a cost-recovery basis. Such an approach could allow governments to concentrate scarce expertise, achieve economies of scale and build stronger professional capability, while creating clearer incentives for HR providers to deliver value to client organisations. By linking funding more directly to demand and performance, this is one way that governments may be able to move HR functions away from a compliance-oriented administrative role and towards a more responsive, customer-focused model that better supports organisational performance and transformation.
3.3.1. Introduction
Governments can have the right priorities, the right people and the right political will and still fail to deliver. What sits between intent and outcome is the operating system of government: the institutional architecture through which decisions are made, the processes through which resources are allocated, the rules through which markets are shaped, and the services through which citizens experience what government actually does. When these systems work well, they translate political commitment into visible change. When they do not, they become a source of the very stickiness this report describes.
The pattern of reform across OECD countries shows governments acting on this. Centres of government are being redesigned to steer, not simply co-ordinate. Institutional architectures are being reviewed for the first time in decades. Procurement is moving from a compliance function toward a strategic one. Regulatory systems are being reoriented toward risk and outcomes. Public financial management is being asked to sustain long-term investment while holding tighter fiscal discipline. And services are being built around the people who use them rather than the institutions that provide them.
These six areas are managed separately in most governments, often by different institutions with little connection between them. But the digital, workforce, and citizen engagement reforms described elsewhere in this report all eventually require significant transformations of the backbone of government, i.e. of its operating systems. Indeed, a government that redesigns its services but cannot procure differently, or that invests in digital infrastructure but has no centre able to steer it, will find itself stuck.
This section examines the institutional architecture and internal processes, centres of government, public procurement, regulation, public financial management and human-centred service design, in turn, and highlights the defining practices that show where the government operating systems are already changing (see Table 3.4).
Defining practices for Agile government operating systems
3.3.2. Fit-for-purpose institutional architecture and tackling beige tape
Key issues
Institutional architecture in OECD countries is shaped by decades of successive decisions. Ministries are created in response to political priorities, agencies are established to deliver specific programmes, arm's-length bodies are set up to insulate operational decisions from political interference, and approval chains and reporting obligations accumulate to manage risk, respond to new mandates and address the lessons of past crises. Each of these decisions makes sense in its own context, but the cumulative effect is an institutional configuration that reflects the layering of past choices as much as it reflects current strategic priorities. Three tensions stand out.
The first concerns the mismatch between inherited structures and the problems governments are now expected to solve. Ministerial architectures organised around historical functions and professional silos struggle to hold cross-cutting priorities such as environmental challenges, managing the demographic transition, digital transformation or economic competitiveness, which do not respect departmental boundaries. A growing number of governments are experimenting with organising institutional architectures around outcomes rather than policy areas, for example through mission-based portfolios, challenge-driven programmes or directional initiatives: the OECD's Mission-Oriented Innovation Policy Online Toolkit tracks 241 active missions across Member countries (OECD, 2026[39]). But restructuring costs more than governments usually expect. Long-standing working relationships break up, informal knowledge goes with them, and capacity can dip while senior officials are absorbed by the reorganisation itself. The question is not whether to restructure but when lighter arrangements, such shared ministerial leadership, or mission-based portfolios laid over existing structures, would serve as well at lower cost.
The second concerns the delivery landscape that has emerged over successive waves of agencification. Separating policy ministries from operational delivery has produced genuine benefits in many OECD countries: greater operational focus, clearer performance accountability, and the ability to attract specialist skills outside standard civil service frameworks. Yet the effect in many systems has been a proliferation of arm's-length entities, each with its own governance arrangements, reporting lines, data systems and back-office functions, generating co-ordination costs and duplicated overhead that consume resources that could otherwise reach frontline delivery. The scale of the reassessment now underway is visible: a third of respondents to the 2026 Restoring Public Finances Survey have merged or abolished agencies in 2025 alone and 46% of OECD countries covered are pursuing shared services and platforms as part of larger cross-government efficiency initiative (see Figure 3.4) (OECD, 2026[40]). For instance, in France, the Mission État Efficace taskforce concluded in July 2026 to a lack of a strategic and whole-of-government approach to deciding when and how to set up, or abolish, agencies.
Measures approved or submitted to parliament for the fiscal years of 2025 and 2026
Note: Results based on 39 Restoring Public Finances Survey responses.
Source: OECD (2026[40]), Restoring Public Finances: Enabling Effective Government, OECD Publishing, Paris, https://doi.org/10.1787/fbcf9161-en.
The third concerns the layers of procedures that fragmented structures generate. Clearance chains, reporting obligations and committee structures are added in response to individual failures, new priorities or accountability crises, but rarely retired when their original purpose has passed. While many internal procedures and rules have been put in place for good reason to mitigate risks or to increase accountability, others yield low marginal benefits compared to their administrative costs. This internal procedural complexity, also known as beige tape (see Box 2.1 of Chapter 2), has no systematic owner, accumulates continuously across political cycles, and remains largely invisible to those outside government. Unlike red tape, which OECD countries have well-developed tools for measuring and reducing, beige tape lacks comparable measurement infrastructure, and the hardest part of reducing it is that no one removes an approval step unless someone is clearly accountable for the decision it was meant to protect.
Defining practices
Some of the most interesting institutional reforms underway across OECD countries are not driven by major restructuring but by more deliberate judgement about when structural change is warranted and how the existing institutional architecture can be made to support more effective delivery. Four practices are emerging that together signal a shift from structural churn toward more considered institutional design.
Prioritising outcomes over structural change. A growing number of governments treat institutional design as a hierarchy of responses rather than a choice between the status quo and full reorganisation. When a mismatch appears, those making most progress first test lighter arrangements, such as joint ministerial ownership, mission-based portfolios, or budgets tied to cross-cutting outcomes rather than administrative units. France 2030 illustrates the ambition: an investment programme of EUR 54 billion directly led by the Prime Minister and delivered through cross-cutting Acceleration Strategies that align the interventions of multiple ministries around shared objectives, without creating a new ministerial structure. The more developed variants add fluidity, with pre-approved legal, budgetary and staffing arrangements allowing taskforces to be stood up and wound down as priorities shift. Where reorganisation is genuinely necessary, experienced practitioners treat it as a delivery challenge in its own right: analysis of options and transition costs beforehand, clear sequencing, interim accountabilities, and political attention beyond the announcement.
Rationalising the agency and delivery landscape. Over one third of countries covered by the 2026 OECD Government Unstuck Survey (15 out of 40 respondents) identify the merging or elimination of ministries, departments or agencies as a significant measure in their current transformation efforts (OECD, 2026[14]). Some governments are conducting systematic reviews of their agency and delivery body landscape to distinguish deliberate design from accumulated habit. This is the case in Germany for example, where the Federal Modernisation Agenda 2025-2028 aims to reduce the number of federal entities from a current estimate of 950, starting with a comprehensive review of their efficiency and effectiveness. Similarly, the United Kingdom’s Public Bodies Review Programme, launched in 2022, to assess whether arm's-length public bodies remained necessary and are operating with effective governance, accountability, efficiency, and alignment with government priorities (UK Cabinet Office, 2022[41]). Where reviews find overlap, consolidation gives citizens a clearer point of contact – and the OECD-European Commission data on citizen satisfaction with public services shows that satisfaction is almost 30 percentage points higher when a user does not have to provide the same information more than once (OECD/European Commission, 2025[42]), suggesting that the case for consolidating is not only about efficiency but about the quality of the service citizens experience.
Treating back-office functions as shared services. Rather than replicating HR, IT, procurement and finance across dozens of entities, governments are consolidating them into shared service centres with clear service-level agreements and structured accountability. This is now mainstream: 46% of countries covered by the 2026 Restoring Public Finances Survey report initiatives to reduce costs through shared platforms and services (OECD, 2026[40]); a similar share of countries (26 out of 40 respondents) covered by the 2026 Government Unstuck Survey report identify shared services or the consolidation of back-office functions as a significant measure in their current efficiency initiatives. Shared services are not a new idea and earlier waves have had a mixed record, but the current wave pays deliberate attention to the co-ordination, quality and accountability difficulties that undermined earlier efforts, and may be largely required by technological developments. Portugal is promoting the creation of shared services for administrative support functions as part of the reform of its Central State Administration, building on earlier experience of shared services in planning and legal advice to inform the design of the current effort. Slovenia has brought 108 authorities into a Unified Work Environment. In several countries, shared services are now treated as the default for new entities, a shift that reflects how difficult it is to migrate arrangements once bodies have built up independent capacity.
Making beige tape visible and giving it an owner. A number of governments are beginning to tackle internal procedural burden explicitly as part of larger transformation efforts. 26 countries out of 40 respondent countries to the 2026 OECD Government Unstuck Survey identify the suppression or reduction of internal regulations and beige tape as a significant measure in their current efficiency initiatives, and 30 countries identify the automation of tasks, including through the use of AI, as central to their efforts (OECD, 2026[14]). What characterises the more interesting practice is the recognition that beige tape accumulates continuously, and one-off reforms tend to be reversed, so simplification needs a standing owner than as a one-off exercise. Estonia has reviewed all 4 009 regulations in the State Gazette, identifying 752 that could be repealed or adapted, including 121 specifically to reduce internal bureaucracy. The United Arab Emirates' Zero Government Bureaucracy Programme, launched in November 2023, similarly illustrates what a systematic and targeted approach to internal simplification can look like at scale. The first cycle set explicit government-wide targets of 2 000 procedures eliminated and processing times cut by at least 50%, as well as mobilised 690 taskforces across 30 federal entities, resulting in the elimination of over 4 100 redundant procedures and more than 1 600 unnecessary requirements. What distinguishes the programme design is the combination of top-down reduction targets with an assessment framework that rewards entities for empowering employees to identify burden, thus anchoring simplification in day-to-day administrative practice rather than treating it as a one-off compliance exercise.
Forward looking issues
Three areas stand out as priorities for further reflection and as directions where the OECD is well positioned to contribute.
Redesigning accountability to make simplification sustainable. The most persistent challenge in tackling beige tape is that approval layers cannot be safely removed unless someone is clearly responsible for the decision that step was put there to safeguard. Without that redesign of accountability, simplification either stalls because officials are reluctant to remove the process that protect them, or works for a time before new steps quietly accumulate to replace the ones that were taken away. The OECD could identify the accountability designs that make simplification durable, document how governments are addressing the risk-aversion dynamics that keep beige tape in place, and build a comparative understanding of the conditions that allow processes to be safely retired (see also the section 3.2 on “Skilled and empowered public servants”).
Building the measurement infrastructure for beige tape. Governments have relatively well-developed tools for measuring red tape imposed on citizens and businesses, but comparable infrastructure for the burden government imposes on itself is largely absent. Metrics such as average clearance times, the number of sign-offs for common actions, or the share of official time spent on internal compliance rather than substantive work remain rare. Without them, reduction targets are hard to enforce, and progress is hard to demonstrate. The OECD could develop shared metrics that capture both the administrative cost and the behavioural frictions of internal process, giving beige tape the same measurement infrastructure that has proved consequential for reducing red tape.
3.3.3. Strategic planning, prioritisation and co-ordination from the centre of government
Key issues
Centres of government have expanded well beyond traditional cabinet support toward more active steering, co-ordination and monitoring of cross-cutting priorities. Three interrelated tensions have sharpened as this role has grown.
The first is scope. Many institutions at the centre of government now carry mandates accumulated over successive administrations, with few incentives to hand issues back once taken on. The result is a centre that is stretched across too many things to steer any of them well, without a clear basis for deciding when to step in, how intensively, and when to step back (OECD, 2026[43]).
The second is the connection between priorities, monitoring and decisions. Political programmes can run to over a hundred commitments, and monitoring platforms have multiplied alongside them. Yet data availability has outpaced data use: monitoring informs without influencing when it sits outside the cabinet calendar, when reporting burdens discourage line-ministry engagement, and when political ownership of the process is absent (OECD, 2026[43]).
The third is the gap between long-horizon analysis and short-term decisions. Foresight remains among the most under-resourced functions at the centre, crowded out by immediate events. A growing number of centres are treating anticipation as a core governing function, but how foresight shapes routine budgeting and resource allocation, as well as planning and preparedness decisions, remains largely unresolved (OECD, 2024[44]; Tõnurist and Orlik, 2025[45]).
Four practices are emerging across OECD countries that together signal a more strategic conception of the centre's role.
Defining practices
Designing the centre's engagement deliberately, including its exit. Some centres now ask, before taking on an issue, whether it requires central ownership or only central convening, and build exit arrangements from the outset. Ireland's handling of the post-2018 policing reform, driven intensively by the Department of the Taoiseach during the implementation phase and then deliberately devolved back to the Department of Justice and An Garda Síochána once line capacity had been built, illustrates how this kind of bounded, time-limited engagement can preserve coherence without overburdening the centre permanently (OECD, 2026[43]).
Matching the co-ordination instrument to the policy problem. Rather than defaulting to permanent inter-ministerial committees, some centres calibrate the response: standing groups for ongoing cross-cutting work, time-limited taskforces for intensive sprints, and dedicated units where co-ordination alone is insufficient. Finland illustrates how this repertoire can be deployed within a single system: four permanent Ministerial Committees anchor co-ordination on enduring domains such as finance, economic policy, EU affairs and foreign and security policy; thematic ministerial working groups focus on a smaller set of government priorities, such as the Ministerial Working Group on Developing the Digital Transformation, the Data Economy and Public Administration, which steers, monitors and implements those priorities; and functional working groups support cross-cutting capabilities such as research and foresight (Government of Finland, 2023[46]). Chile’s State Transformation Agenda 2026-2030 uses a layered and flexible co-ordination framework. A State Modernisation Committee of senior cross-government decision-makers provides strategic steering and helps resolve cross-institutional bottlenecks, while an independent Advisory Council and thematic technical committees provide specialised expertise where needed to steer and monitor the Agenda’s priorities.
Monitoring governing priorities and reviewing them actively. Some centres take a more active role in narrowing the political programme to a smaller set of clearly assigned, time-bound priorities with named leads and measurable outcomes. More importantly, they own a recurring process to monitor, prioritise and deprioritise, consolidate or rescope commitments as realities evolve. Latvia illustrates how this can be built into institutional routine, with a state policy planning framework that connects long-term development plans and government action plans in a single system, and an internal Government Project Monitoring Office in the Prime Minister's Office tracking selected priorities on behalf of the Central Council, designed to preserve continuity of long-term direction while allowing incoming governments to revise adopted plans as political priorities shift. This kind of active curation, supported by monitoring that reaches the moments where decisions are made, is what protects the strategic bandwidth that delivery requires (OECD, 2026[43]).
Turning foresight into decisions and preparedness. Advanced centres are closing the gap between long-horizon analysis and short-term operations by connecting foresight to budgeting, resource allocation and risk preparedness. Finland illustrates the long-horizon end of the practice, with a shared foresight method embedded across ministries over more than two decades, anchored by a Government Report on the Future submitted each electoral term and scrutinised by a standing parliamentary Committee for the Future, a structure designed to feed foresight into budget and planning cycles rather than parallel strategic documents. National risk assessments such as the United Kingdom's National Security Risk Assessment sit at the interface, translating long-term signals into a prioritised view of the risks government must be ready for; and real-time situational awareness infrastructure such as the United Kingdom's National Situation Centre and Australia's National Situation Room increasingly draws on foresight and modelling to stress-test assumptions and surface emerging risks before they escalate. Ireland's approach illustrates a similar two-part model at smaller scale: a recurring National Risk Assessment provides the horizon-scanning function, while the Department of Finance's Future Forty, a fiscal and economic outlook to 2065, extends the planning horizon well beyond the electoral cycle, linking long-term demographic and economic scenarios directly to fiscal strategy.
Forward looking issues
The defining practices set out above show that some centres are moving deliberately toward a more strategic conception of their role. Yet there are areas where interesting possibilities are visible but convincing country practices are either thin or underexplored, or where the transformation underway raises questions that no jurisdiction has yet fully answered. Three areas stand out as priorities for further exploration.
Making the connection between priorities and budgets real. Without a link to budget allocations, priorities remain strategic statements rather than governing commitments. Few countries have built an institutional routine that consistently delivers this. The OECD is well placed to map how existing practices in this space actually function, identify the institutional conditions that make them consequential, and develop guidance on how CoGs and the Ministry of Finance can work together without either capturing the other.
Harnessing AI and real-time analytics for centre-level steering. The gap between data availability and data use is one of the more persistent findings across OECD work on centres of government, and AI-supported analytics (for example, automated synthesis of progress data across ministries, early warning flags for priorities at risk, natural language summaries tailored to decision makers, real-time cross-referencing of administrative datasets) offer genuine potential to close it. What is less clear is how these tools can be embedded in centres in ways that strengthen rather than displace the political judgement, cross-government convening and institutional memory that give the centre its distinctive contribution. The OECD could support countries by documenting emerging use cases across CoGs, developing shared standards for the responsible use of AI in centre-level steering, and drawing lessons from early adopters about what works, what fails, and what the institutional preconditions are for the technology to add value.
Centres of government as coalition builders for reform. Some of the most significant recent transformations in government, such as the COVID-19 response, energy transition planning, major digital reforms, have depended not only on cross-ministerial co-ordination but on the centre's ability to build broader coalitions for change: with parliaments, subnational governments, social partners, civil society organisations, the academic community, and, at times, private sector actors whose co-operation is essential to delivery. Yet the centre's role as a coalition builder is under-conceptualised compared to its role as a co-ordinator inside government. What are the institutional arrangements that let centres engage credibly with actors outside government without diluting the political mandate of elected ministers or infringing on the autonomy of subnational government? How can the convening power exercised during crises be sustained as a routine capability? The OECD could contribute to this reflection by documenting how centres have exercised this convening role in recent reform episodes, including the COVID-19 response, and by developing a comparative understanding of the institutional structures, advisory councils, standing dialogue forums, cross-sectoral taskforces, through which some centres are beginning to institutionalise coalition building.
Together, these three areas point to what may be the defining challenge for the next generation of central steering. The OECD's continuing work with centres of government offers a platform for exploring these questions comparatively, and for surfacing the practices that will define this next phase.
3.3.4. Public procurement: from compliance to strategy and simplification
Key issues
Public procurement represents 12.7% of GDP across OECD countries, making it one of the largest and most consequential functions performed by governments (OECD, 2025[47]). Governments are not only among the largest buyers in their economies; through the choices they make about what to buy, from whom and on what terms, they shape markets, influence innovation, and translate political priorities into concrete outcomes for citizens and businesses.
At the same time, in many OECD countries, procurement has historically been a rules-based, compliance-driven function, with a primary focus on ensuring the proper use of public funds. This approach has placed strong emphasis on transparency, integrity, access, balance, participation, efficiency, e-procurement, capacity, evaluation, risk management, accountability and integration, all of which are reflected in the twelve principles of the OECD Recommendation on Public Procurement and remain foundational to well-functioning procurement systems (OECD, 2015[48]). However, as governments face rising pressures on public finances, growing expectations that procurement should contribute to strategic policy objectives, as well as rapidly changing markets and technologies, what constitutes fit-for-purpose procurement have also evolved. Three interrelated tensions are emerging across OECD countries.
The first concerns the gap between the strategic ambitions for public procurement and the culture and systems that must deliver them. Governments increasingly expect procurement to advance a growing list of objectives: environmental sustainability, innovation, support to Small and Medium Enterprises (SMEs), resilience of supply chains, economic security, industrial development and social outcomes. The 2024 OECD Public Procurement Survey shows that all responding countries integrate at least one strategic policy objective into procurement policy documents, with environmental objectives cited by 100% of country respondents, innovation by 95% and support to SMEs by 90% (OECD, 2024[49]). In practice, procurement in many OECD countries remains culturally, institutionally and regulatorily oriented around compliance and price rather than outcomes and public value. This shapes what public buyers prioritise, the risks they take, and the signals governments send to the market. The gap between strategic ambition and operational reality is compounded by the persistent absence of impact measurement: while 85% of OECD respondents define Key Performance Indicators for their procurement systems, only 48% measure the impact of procurement on administrative efficiency, 23% on the economy, 10% on environmental and social outcomes, and 8% on trust in government (OECD, 2024[49]). Without robust measurement frameworks, governments have little basis for assessing whether strategic procurement is achieving its intended objectives, managing trade-offs between competing priorities, and holding the procurement system accountable for anything beyond procedural compliance.
The second concerns a set of tensions between balancing efficiency, access and innovation, which can be difficult to navigate within existing procurement systems. Consolidation tools such as central purchasing bodies, framework agreements and dynamic purchasing systems can deliver economies of scale and reduce transaction costs, but only if their design does not simultaneously narrow supplier bases and shut out the firms that would broaden market participation. As of 2024, 93% of OECD countries had established national or federal central purchasing bodies for general commodities, but specialised bodies remain far less developed, with only 50% using them for health procurement and 30% for ICT (OECD, 2025[50]). As a result, the benefits of consolidation usually remain concentrated in categories where specialised expertise is less critical. Access to procurement opportunities is under similar strain from a different direction. Although procurement systems are formally open to any qualified supplier, they often remain difficult to access for firms lacking the scale, legal expertise or administrative capacity to navigate procurement procedures. SMEs, which represent approximately 99% of firms and generate 50 to 60% of value added across OECD economies, continue to face structural barriers related to administrative complexity, financial capacity and the size of contracts. Indeed, 55% of European SMEs identify regulatory and administrative burden as their greatest challenge, whereas foreign and remote suppliers face parallel constraints, with 45% of firms citing burdensome administrative requirements and 41% citing difficult access to information. These two dynamics — consolidation that narrows the supplier base, and administrative barriers that keep new entrants out — reinforce one another, and both bear most heavily on the firms most likely to bring new solutions into the procurement market. Innovation-oriented instruments such as competitive dialogue, innovation partnerships and challenge-based approaches are designed precisely to counter this: by selecting suppliers on the solution proposed rather than on scale or track record, they can widen the supplier base at the same time as they source new capability. Yet innovation procurement accounts for only 10% of procurement expenditure in the European Union (Draghi, 2024[51]), reflecting a deeper incentive problem: public buyers are rewarded for avoiding process errors and penalised for visible failures, which discourages the departure from standard specifications that these approaches require. The instrument best placed to reconcile efficiency, access and innovation is therefore also the least used.
The third and final challenge concerns the digital foundations on which strategic, simplified and open procurement now depends. Digital technologies are no longer simply tools to support procurement processes; they have become the infrastructure on which modern procurement systems operate. As such, 98% of OECD countries use a central online platform, and 70% report using emerging technologies, including artificial intelligence and cloud, as part of the procurement system (OECD, 2025[50]). Yet in many countries, digital transformation has focused primarily on digitising analogue procedures rather than fundamentally redesigning them. Fully integrated end-to-end e-procurement systems, from needs assessment through tendering, contract management and payment, have been shown to increase bidder participation, reduce single-bid procedures and deliver better value for money. As AI and data analytics move from experimentation to mainstream use, and as procurement systems are increasingly expected to interoperate with public financial management, tax and business databases, the challenge is no longer whether to digitise procurement but how to design digital systems that foster strategic, efficient and open procurement, and enable a shift from process-based compliance towards outcome-based procurement.
Defining practices
Across OECD countries, some of the most significant procurement reforms are not driven by drastic changes to the regulatory framework itself, but by practical changes in how the function is managed, resourced, measured, accessed and digitised. Four practices are emerging that together signal a shift from a compliance-centred conception of procurement towards one that places greater emphasis on the strategic outcomes it delivers, while preserving the integrity, transparency and access principles on which the legitimacy of the function ultimately rests.
Anchoring strategic procurement in national priorities and measuring what it actually delivers. A growing number of governments are moving beyond generic statements of strategic intent toward explicit integration of procurement into national strategic frameworks, with measurable, enforceable targets attached and central oversight to prevent public buyers from pursuing divergent objectives through the same procurement system. Mexico's Plan Mexico, the country's economic strategy for equitable and sustainable development and shared prosperity, includes as its ninth action a target of 65% national content in goods and services procured by the federal government, supported by a reform of the public procurement regulatory framework in April 2025 (Federal Government of Mexico, 2025[52]). New Zealand's Government Procurement Rules require the integration of environmental considerations across all levels of government and for all contracts meeting or exceeding the NZD 100 000 value threshold, giving suppliers a consistent signal and public buyers a common frame. The more interesting shift underway, however, concerns what happens after strategic priorities are set: a small but growing number of countries are beginning to measure whether strategic procurement is actually delivering the outcomes it was designed to advance (OECD, 2025[50]). France is tracking employment outcomes generated through social inclusion clauses (Government of France, 2025[53]; 2025[54]), and Colombia has used procurement data to conduct differential analysis of gender participation and salary gaps among public sector contractors, drawing on data from its e-procurement platform SECOP to identify where women's participation is growing, where gaps persist, and how the composition of the supplier base varies across sectors (OECD, 2023[55]). What characterises these practices is a recognition that anchoring strategic policy objectives at the centre and measuring strategic procurement outcomes are interdependent: without the first, strategic ambition is not translated into consistent policies; without the second, governments cannot know whether the ambition is being met. These efforts do not displace the transparency and access commitments set out in the OECD Recommendation on Public Procurement, which remain the foundation on which strategic procurement is built (see Box 3.4).
The OECD Recommendation on Public Procurement provides a holistic vision for a well-functioning public procurement system through its twelve integrated principles: Transparency, integrity, access, balance, participation, efficiency e-procurement, capacity, evaluation, risk management, accountability and integration.
While the principle on “Balance” recognised the strategic use of public procurement to achieve specific public interest goals or policy objectives. The principle on “Access”, calls countries to facilitate access to procurement opportunities for potential competitors of all sizes through different means, including by:
treating bidders, including foreign suppliers, in a fair, transparent and equitable manner, taking into account Adherents’ international commitments
ensuring that specific tender opportunities are designed so as to encourage broad participation from potential competitors, including new entrants and small and medium enterprises
The principle on “Transparency” calls countries to ensure an adequate degree of transparency of the public procurement system in all stages of the procurement cycle, including by promoting fair and equitable treatment for potential suppliers by providing an adequate and timely degree of transparency in each phase of the public procurement cycle.
Source: OECD (2015[48]), “Recommendation of the Council on Public Procurement”, OECD Legal Instruments, OECD, Paris, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0411.
Leveraging efficiency tools to maximise value-for-money. With public finances under sustained pressure from rising debt servicing costs, ageing populations and new spending demands, generating value for money from the 13% of GDP that flows through public procurement has become a central concern for OECD countries, and efficiency tools are among the most direct levers available. Beyond the widespread use of general central purchasing bodies (93% of OECD countries), a growing number of governments are pursuing efficiency gains through more specialised or collaborative arrangements. Specialised central purchasing bodies (CPBs) focused on health, ICT and defence, where market knowledge and technical skills matter most, remain far less developed than general ones, but where they exist, they concentrate expertise and generate value in categories that generic purchasing struggles to serve. In France, for example, the CPB Resah is dedicated to health procurement with purchasing categories such as healthcare products, biology, and patient care, and related services, while in Hungary, Digital Government Agency centralises the government’s purchases in the ICT sector (OECD, 2025[50]). Collaborative tools such as joint procurement between public buyers, framework agreements and dynamic purchasing systems also allow demand to be consolidated without full centralisation. Some countries are extending these approaches to the sub-national level, where around two-thirds of public procurement takes place and where individual authorities are often too small to procure efficiently on their own. In Slovenia, for example, the municipality of Ljubljana acts as a leading entity by providing contracting authorities within Ljubljana with joint procurement opportunities as well as capability-building opportunities. For this purpose, Ljubljana municipality collects annual needs, and procurement plans of contracting authorities in order to group their needs and leads procurement procedures on their behalf (OECD, 2025[56]). Finally, simplification helps alleviate administrative burden for both public buyers and economic operators. For instance, France has lowered the minimum revenue required of companies bidding on public contracts, and Latvia is aiming to shorten and reduce procurement procedures with expected national budget savings of 2-4% (OECD, 2026[40]).
Opening the public procurement market to a broader and more dynamic supplier base. A growing number of governments are approaching supplier participation as a design question rather than a residual outcome, deploying concrete tools to lower structural barriers, sustain participation over time, and actively engage suppliers who would not otherwise bid. Measures taken by governments include dividing contracts into lots to match SME capacity, simplifying financial guarantees, using functional rather than prescriptive technical specifications, and designing framework agreements and dynamic purchasing systems with SME participation in mind. Denmark's central purchasing body, SKI, analyses the structure of supply markets and the potential role of SMEs when designing tenders, and adapts its procurement strategy accordingly (OECD, 2024[57]); France's bourse à la cotraitance on the State e-Procurement platform PLACE helps firms form temporary consortia to bid for contracts they could not pursue individually (OECD, 2025[58]). Sustaining participation over time also depends on how contracts are managed after award: Korea's Public Procurement Service provides advance payments of up to 70% of contract value to SMEs and pays invoices for certain goods contracts within hours. Where market conditions justify it, some governments use set-asides or participation targets: Japan reserves a substantial share of procurement expenditure for SMEs, and Australia has committed to procure 25% of contracts up to AUD 1 billion and 40% of contracts up to AUD 20 million from SMEs. Innovation procurement is supported in Austria and Croatia by dedicated competence centres that provide public buyers and economic operators with the guidance, tools and financial incentives a compliance-oriented culture would otherwise withhold. Active supplier engagement is also being institutionalised: Colombia's Colombia Compra Eficiente uses working groups with suppliers, manufacturers, distributors and buyers from ICT markets to establish benchmarks and develop tender documents before tendering begins (OECD, 2026[59]).
Digitalising procurement “end-to-end” for radical simplification. Digital transformation involves rethinking the entire public procurement lifecycle, integrating digital technologies into every phase, and embedding a digital-first mindset across public institutions. For instance, Korea's ON-line E-Procurement System (KONEPS), launched in 2002 and now interconnected with 227 external systems from both the public and private sectors, illustrates what a mature end-to-end approach looks like. Through this interconnection, KONEPS automatically retrieves the documents and certifications suppliers would otherwise have to submit at each stage of bidding and contracting, such as business registration certificates, tax payment certificates, letters of guarantee, qualification screening documents and certified copies of corporate registers, directly from the government and financial platforms that hold them, meaning that suppliers do not need to re-enter or re-submit information the state already has. This radically reduces the administrative burden on both suppliers and public bodies, and applies the once-only principle at the heart of the procurement lifecycle rather than only at its edges. The Next-Generation KONEPS, launched in March 2025, uses AI and Big Data to predict bidding congestion, monitor procurement operations and recommend products, with dedicated support for SME participation (OECD, 2025[50]). AI and data analytics are also increasingly being used to support decision making across the procurement lifecycle. Portugal's Court of Auditors has developed a data-driven risk assessment methodology for audit selection in procurement, drawing on advanced analytics and AI to focus limited oversight resources where they have the greatest impact (OECD, 2024[60]). Beyond compliance, data-based analytics and AI can play a role in areas such as strategic market analysis and contract management, including by identifying potential new entrants, benchmarking prices and monitoring contract execution against agreed milestones and outcomes.
Forward looking issues
The defining practices set out above show that some governments are moving deliberately toward a more strategic conception of public procurement. Yet the frontier remains open, and there are areas where interesting possibilities are visible but convincing country practice is still thin, or where the transformation underway raises questions that no jurisdiction has yet fully resolved. Three areas stand out as priorities for further exploration, both for procurement systems themselves and for the OECD's future work in supporting them.
Balancing competing objectives while ensuring sound competition. Procurement systems are increasingly expected to advance several objectives simultaneously, including fostering innovation, achieving efficiency, supporting SME’s development and, more recently, advancing industrial policy and resilience goals. While these objectives can be mutually reinforcing, they may also create tensions, particularly when governments seek to balance open and competitive procurement markets with strategic priorities. Within industrial policy, some countries are increasingly using domestic-content requirements and award preferences for domestic companies. In parallel, countries are increasingly developing reciprocity instruments and subsidy controls to support “fair competition”, marking a shift toward conditional access to procurement markets rather than open market access. What remains uncertain is whether these trade-offs can be managed explicitly rather than implicitly, and how to distinguish procurement interventions that genuinely advance public policy goals and deliver value for money from those that mainly shields domestic firms from competition. A related challenge is how governments prioritise among competing objectives when trade-offs arise. As procurement is increasingly expected to contribute simultaneously to innovation, resilience, SME participation, industrial development and value for money, few countries have developed systematic methodologies to determine which objectives should prevail in specific contexts, sectors or procurement categories. Governments may also need mechanisms for establishing a hierarchy of objectives or for defining the circumstances under which certain objectives should take precedence over others. Likewise structured approaches to assess the economic, trade, resilience and market impacts of procurement-based interventions remain limited, and the analytical tools are still underdeveloped (OECD, 2026[61]). To help governments manage these tensions, the OECD could develop analytical frameworks for assessing the trade-offs and prioritising objectives, beginning with an update of its 2017 Taxonomy of Measures Affecting Trade in Government Procurement Processes to capture newer explicit and procedural measures that can affect foreign suppliers’ access to public procurement. It could also develop impact-assessment methodologies, including cost-benefit analysis that identify who bears the costs of a measure, assess the long-term implications, enabling procurement interventions to be evaluated on evidence rather than simply asserted.
Advancing from e-procurement to integrated digital public procurement: Beyond the technological frontier illustrated by systems such as KONEPS, the wider challenge for most OECD countries is organisational rather than technological. Moving from e-procurement to integrated digital procurement depends on interoperability standards, governance arrangements and, critically, whether procurement workforces have the skills to operate in digital, data-rich and strategically oriented environments. The OECD could support countries by identifying good practices and governance models for digital transformation, developing guidance on the use of AI in procurement, and examining the interoperability and institutional conditions required for greater integration with wider digital government systems.
Turning procurement data into policymaking infrastructure. Digital transformation is generating unprecedented volumes of procurement data, yet in most countries these data remain fragmented across systems, are of low quality and are primarily used for compliance and reporting rather than policymaking. Furthermore, data does not always cover the whole public procurement cycle, from planning to the conclusion of the contract, and how well the contractual cycle supports the policy outcomes that public purchasing is meant to achieve. Transforming procurement data into a strategic asset that informs policy and operational decisions requires moving beyond standalone e-procurement platforms towards integrated ecosystems that connect with wider government data, including public financial management, business registers, tax systems and contract performance databases. This integration can support more informed needs analysis, stronger commitment control, real-time monitoring of budget execution, and better contract management. For example, shared budget classifications and real-time data flows allow analytics and artificial intelligence to operate in ways no single system permits: detecting fraud and collusion as transactions occur, flagging emerging budget overruns, and revealing the full cost of a service across labour, procured inputs and assets. However, relatively few governments have established the governance arrangements, common standards and analytical capabilities to translate data into actionable intelligence. The OECD could help countries develop end-to-end public procurement data strategies to integrate with broader government data ecosystems and reposition procurement data as a source of operational, fiscal and strategic insight.
3.3.5. Risk-based and future-proof regulations
Key issues
Regulatory systems across OECD countries are converging on a shared ambition: that regulation should be proportionate to risk, grounded in evidence, and able to keep pace with change. In other words, governments generally seek to maximise the public value generated by regulation, while minimising costs and complexity, generally avoiding the extremes of extremely limited regulatory frameworks or overly complex and prohibitive ones (as stylised in Figure 3.5). Simpler regulation is one of the elements the direction of change describes, and it is largely delivered, or not, through the systems set out here. Yet the distance between that ambition and current practice remains wide. Three tensions are coming into focus.
Source: OECD (2026[62]), Smart Regulations, Strong Business: Enabling Growth and Societal Protection in a Fast‑Changing World, OECD Publishing, Paris, https://doi.org/10.1787/93d38770-en.
Regulation is not consistently calibrated to risk. Effective risk-based regulation begins with understanding where risks lie and how best to address them. This depends on collecting the data needed for risk analysis and then performing a robust risk assessment, which together provide the foundation for the decisions that follow: which regulatory tools to use or to avoid, how intensively to intervene, and where to direct limited regulatory resources. Yet the policy foundations for such an approach remain only partial: only around one-third of OECD Members require risk-based enforcement, and 24 of 38 have no policy basis for differentiated, responsive enforcement (OECD, 2025[63]). Proportionality, moreover, is not only a question of how intensively to regulate but of who should regulate. A proportionate response depends on the maturity and risk profile of the activity: serious risks may justify command-and-control measures such as mandatory standards or bans even at an early stage, whereas lower-risk or fast-evolving activities may be better served by performance-based, co-regulatory or industry-led arrangements that preserve room to innovate. The difficulty is compounded by institutional incentives. Regulators frequently default to "gold-plating" because they bear the reputational cost of a failure but not the cost of the compliance burden they impose, as noted in the OECD report on Accelerating Infrastructure Permitting (forthcoming[64]) The question this raises is whether systems can calibrate both the intensity of regulation and the balance of public and private roles to the risk actually at stake.
Rulemaking remains anchored in process rather than outcomes. Data from the Regulatory Policy Outlook suggest that ex ante impact assessment and consultation often function as "tick-the-box" exercises, conducted late in the cycle around an already-preferred option. In roughly 70% of countries, stakeholder engagement takes place only after a proposal has been developed, and cost-benefit analysis quantifies more than one policy option in fewer than half of OECD Members (OECD, 2025[63]). Implementation questions are frequently deferred to after the design stage, at measurable cost. For example, the OECD report on Accelerating Infrastructure Permitting (forthcoming) records 30% real cost growth in Swedish transport projects between the formal decision to build and tendering. Cost growth of that kind falls on public budgets, and is among the quieter drains on the margins of manoeuvre the direction of change seeks to restore. Compounding this, engagement tends to amplify the best-resourced voices while SMEs, dispersed citizens and diffuse public benefits are systematically under-weighted. The open question is whether rulemaking can be reorganised so that evidence, alternatives and implementation genuinely shape decisions before a preferred option hardens.
Regulatory institutions struggle to keep pace with fast-moving, interconnected challenges. Established institutions were not designed for the speed, complexity and interconnectedness of contemporary policy problems, and digital technologies sharpen the difficulty: they evolve rapidly, cut across sectoral mandates, and are deployed across borders. Fewer than half of OECD countries use horizon scanning systematically for regulatory innovation, and only 9 of 39 have co-ordination arrangements to identify cross-cutting innovation issues (OECD, 2025[63]). Keeping pace depends heavily on institutional readiness with the support of the expertise, digital tools and co-ordination mechanisms available to regulators, yet more than half of surveyed countries report difficulty recruiting the digital and data expertise that effective oversight requires (OECD, 2022[65]), and only 41% of citizens consider it likely that their government will regulate new technologies appropriately (OECD, 2024[66]). Being prepared in a digital world includes being able to regulate what that world produces. The open question is whether regulatory systems can become genuinely anticipatory and adaptive and equip themselves with the skills and tools to steer private action, without sacrificing the predictability on which regulated entities depend.
Defining practices
Some governments are moving deliberately to close these gaps, and their approaches point to what a risk-based and future-proof regulatory system can look like in practice.
Actively managing the stock of regulations: First, there will be a need to simplify the stock of regulations, which will require engaging strategically with stakeholders on where and how to simplify in a staged and defined manner. Starting with broad consultations to determine priorities for consultation, or gather ‘common sense’ simplification ideas from business and citizens like in France 2023 consultations is a solid start. More targeted consultations on simplification need to follow and keep the simplification engagement in the long run such as in Spain with the annual Convenios para la identificación de cargas administrativas y su reducción. Other long-term efforts such as in Switzerland with the Federal Act on Reducing Regulatory Costs for Businesses, adopted in 2023, which requires the Federal Council to identify three to five priority areas each year in which they will examine opportunities for burden reduction will also help continue the efforts on burden reduction over the long term.
Calibrating licensing, inspection and enforcement to risk. Some countries are reserving licensing and permitting for high-risk activities and applying alternative instruments, such as notifications with assessment, simple notifications, or no instrument at all, as risk descends, structuring the choice as a risk pyramid (OECD, 2025[67]). They are formally requiring regulators to base inspection planning on risk criteria and consolidating overlapping inspections across authorities, as well as shifting from punitive to graduated, responsive enforcement that prioritises correction and compliance promotion before sanction. Australia's Online Safety Amendment (Social Media Minimum Age) Act, in force since December 2025, illustrates an outcomes-based approach: it requires age-restricted platforms to take "reasonable steps" to prevent under-16s from holding accounts, leaving them flexibility on how to comply, while the eSafety Commissioner supports compliance through published guidance, a platform self-assessment tool, and a regularly updated list of in-scope services and issues information notices to inform assessments of compliance.
Moving stakeholder engagement further upstream and closing the loop. Practitioners are consulting on the problem and the options before a preferred regulatory draft exists, and sequencing engagement across the cycle: broad consultation at priority-setting (e.g. France's 2023 consultation “Conseil National de la Refondation” to shape the policy agenda before specific proposals were developed), targeted engagement at solution design (e.g. Spain's Convenios (collaboration agreements) were used to engage stakeholders at the solution-design stag), and focused technical consultation on draft texts last. The European Commission's Call for Evidence invites feedback on the issue, objectives and policy options before a draft regulation is developed, as do the United Kingdom's pre-draft online consultations. Accountability is reinforced by "you said / we did" responses: on Canada's Let's Talk Federal Regulations platform, policymakers respond directly to comments, and the United Kingdom routinely issues a consultation response, typically within 12 weeks.
Anchoring impact assessment in outcomes and testing them proportionately. Some countries are designing decision making systems to be geared towards the outcome to be achieved and genuinely weigh the alternatives for reaching it, including non-regulatory ones, before settling on an instrument, rather than assessing a single preferred option after the fact. They also scale the depth of that appraisal to what is at stake: Canada applies RIA to all subordinate regulations but uses a Triage System, anchored in the Cabinet Directive's proportionality principle, to determine early whether a proposal needs full or expedited analysis, while the United Kingdom's formal preliminary assessment sorts proposals into low, medium and high impact bands against defined net-cost thresholds. The UK also strengthens substantive challenge: an independent Regulatory Policy Committee must give an opinion on the options assessment, covering the rationale for intervention and the justification of the preferred option, before that option proceeds to collective agreement.
Matching regulatory approaches to risk, maturity and market conditions. Rather than defaulting to prescriptive rules, some governments are choosing deliberately along a spectrum of governance models (see Box 3.5 for a description of this spectrum), blending public and private roles to fit the technology and market at hand, and often combining several models in a single framework, as the EU does in the AI Act. Where risks are serious and public understanding is mature, command-and-control retains its place, as in the EU's mandatory USB-C common-charging standard. Where measurable outcomes are hard to specify, principle-based approaches supported by guidance are used instead: Singapore's Personal Data Protection Act sets general obligations and the data protection authority issues advisory guidelines, including a self-assessment annex, to steer compliance and limit divergent interpretation. Co-regulation lets public actors set the frame while industry and civil society supply the detail, as in the EU Digital Services Act and its Strengthened Code of Practice on Disinformation, or the embedding of Data Protection Officers as enforcement actors under the GDPR. Market-based mechanisms embed objectives directly in design, as in Chile's 2021 5G spectrum auction, which built coverage, bandwidth and pro-competition caps into the auction rules. And where an industry lacks a self-regulatory body, public actors can construct the framework for one: Australia's Online Safety Act empowers the eSafety Commissioner to require industry to co-develop Online Safety Codes regulating illegal and age-restricted material and, where codes fail to provide appropriate community safeguards, determine mandatory Standards directly, while Canada's and Australia's news-media bargaining codes use a public arbitration backstop to structure private agreements.
Governments regulating the digital economy face a question that goes beyond whether to intervene: how to distribute responsibility between public and private actors in a way that is proportionate to the risk, suited to the maturity of the technology, and realistic about institutional capacity on both sides.
The OECD working paper on The Spectrum of Regulatory Models and Its Application to the Digital Economy (King et al., 2026[68]) maps six established models along a spectrum from strong public control to industry self-regulation: command-and-control, performance-based regulation, market-based mechanisms, co-regulation, meta self-regulation and industry self-regulation.
Governments rarely adopt a model whole. The EU AI Act, for instance, combines elements of at least four models in a single framework. The paper finds that across the digital economy, governments increasingly retain responsibility for setting regulatory objectives while delegating design and delivery to private actors under public oversight, a pattern that reflects both the technical expertise industry holds and the information asymmetries that make traditional inspection difficult. It also finds that the two ends of the spectrum, pure command-and-control and pure industry self-regulation, are the least likely to be used for digital services, since neither fits the pace, complexity and cross-sectoral character of the technology well.
Three criteria guide model choice: the risk profile and maturity of the technology being regulated; the willingness, capacity and market structure of the private actors involved; and the institutional readiness of the public authorities, including their skills, digital tools and co-ordination arrangements.
The paper concludes that the task for regulators is less about identifying a single right model than about building the institutional conditions to choose well, adjust when circumstances change, and sustain the legitimacy of regulatory action as technologies and markets evolve.
Source: King, B. et al. (2026[68]), “The spectrum of regulatory models and its application to the digital economy”, OECD Regulatory Policy Working Papers, No. 26, OECD Publishing, Paris, https://doi.org/10.1787/b7456763-en.
Designing regulation for implementation and adaptation. Some governments now settle enforcement, data and guidance questions at the design stage rather than deferring them. Denmark has, since 2018, assessed new laws against seven digital-ready legislation principles, screening over 800 laws and 300 proposals to date; while Germany's Digitalcheck maps legislative processes with flowcharts and decision trees to surface conflicting rules, redundant requirements and digital-implementation obstacles before laws take effect. Implementation is then treated as a continuing task rather than a one-off. Threshold-based ex post review obligations are baked into legal frameworks, and governments increasingly treat the period before and after intervention as active: monitoring markets through horizon scanning, foresight and observatories, and revisiting governance choices as technologies mature. Bespoke advisory bodies such as the UK's Regulatory Horizons Council channel weak signals into anticipatory regulatory governance. Ongoing supervision keeps private-led arrangements honest, as with the Australian Competition and Consumer Commission's Digital Platform Services Inquiry, sustained since 2017 "in the shadow" of potential public intervention. Some regulators now run sandboxes as a permanent function, letting innovators test products in live but bounded conditions while the regulator learns what the rules should be: the Norwegian Data Protection Authority's sandbox supports compliance while sharpening its own supervisory methods and Spain's AI sandbox turned pre-application testing of the EU AI Act into guidance for all regulated entities.
Forward looking issues
The defining practices set out above show governments moving deliberately toward more risk-based and adaptive regulation. Yet several areas remain where opportunities are visible, but country practice is still thin, or where the transformation underway raises questions no jurisdiction has fully resolved, and where the OECD is well placed to support the reflection.
Addressing the misaligned incentives that distort regulatory decisions. Proportionality depends not only on the rules a system adopts but on whether the incentives of those designing and implementing them are aligned, and these incentives sometimes pull in unhelpful directions on both sides. Regulators gold-plate because they bear the reputational cost of failure but not the compliance cost of the measures they impose. In fact, findings from the 2025 OECD Simplifying for Success (S4S) survey highlight that reporting costs are the most widely cited source of burden, identified by business organisations in 67% of countries, ahead of compliance costs associated with substantive requirements (53%) (OECD, 2026[62]). On the other side, stakeholders with bargaining leverage at operational or technical stages of decision making can hold up or reshape outcomes, with the best-resourced prevailing over diffuse interests. What remains uncertain is how to correct either dynamic: few countries have deliberately redesigned the performance measures and accountability arrangements that reward regulators for minimising rather than managing risk, and structured, rights-based approaches to bargaining, transparent and time-bound negotiation with independent valuation and pre-committed compensation, confined to the decisions where bargaining genuinely belongs, remain rare and largely untested. The OECD could investigate how these incentives are currently structured, document the few cases where regulator incentives have been reoriented toward active risk management or where bargaining has been deliberately channeled and contained, and build the comparative evidence needed to make both defensible, including through the OECD S4S Initiative (see Box 3.6).
The initiative
The OECD Simplifying for Success (S4S) initiative supports governments in developing simple, smart and streamlined regulatory frameworks that help unlock prosperity while safeguarding the public interest. Building on the OECD’s long-standing work on regulatory policy and governance, the initiative responds to growing concerns that increasingly complex regulatory environments can undermine productivity, innovation and the effective delivery of policy objectives. The S4S initiative provides a platform for governments, businesses, consumer organisations and academics to exchange experiences and identify practical approaches to reducing unnecessary regulatory burdens, simplifying administrative procedures and improving the quality of regulation, while maintaining high standards of protection for citizens, workers and the environment. The OECD High-Level Symposium “Simplifying for Success: Smart Rules, Stronger Business”, held in Paris on 17-18 November 2025, provided a key opportunity for senior representatives from government, business, trade unions, consumer organisations and academia to discuss lessons from past simplification efforts and emerging approaches to tackling regulatory complexity.
Key findings from the 2025 OECD S4S Surveys
The OECD S4S Surveys, conducted between July and September 2025, gathered information on perceptions of regulatory burdens, ongoing simplification initiatives and priorities for future reform across:
Government respondents from 28 OECD Member countries, 6 Accession countries and the European Union.
Business and employers’ organisations that are national member and observer organisations of Business at OECD (BIAC), across 27 OECD Member countries and 4 Accession countries.
The survey evidence contributed to the development of the OECD report Smart Regulations, Strong Business: Enabling Growth and Societal Protection in a Fast-Changing World, which examines how governments can address both the symptoms and underlying causes of excessive regulatory complexity. The report highlights that:
72% of government respondents, and business organisations in 90% of countries, agree that regulation and bureaucracy are excessive.
in 77% of countries, business organisations say administrative compliance costs increased in the past three years.
89% of government respondents say past burden reduction and simplification efforts were partly successful at best.
Capturing the benefits and wider effects of regulation, not only its administrative burden. The dominant measurement tools, such as the Standard Cost Model, quantify administrative burden narrowly, leaving the benefits of regulation and its dynamic and second-round economic effects largely outside the evidence base, as are the costs and benefits of aligning with, or diverging from, international standards. The open question is whether these wider effects can be measured robustly enough to inform decisions before an instrument is chosen. Promising but nascent tools point the way, including the Australian Social Value Bank, RegValue, the EESC's AI-supported Regulatory Burden Index, and European Commission work assigning a market value to positive regulatory outcomes, but none is yet established practice. The OECD could develop and help standardise methodologies that capture benefits and indirect effects alongside costs, giving impact assessment a more balanced evidentiary foundation and reinforcing the outcomes-based appraisal that leading systems are beginning to adopt.
Turning rule-as-code and AI into genuine enablers of simpler regulation. Publishing law in consistent, machine-executable formats could underpin a step change in simplification, from AI-enabled quantification of regulatory burden to assisted legal drafting, and, crucially, could provide accurate and verifiable outputs where purely AI-based systems remain probabilistic and prone to hallucination. What remains uncertain is how far this can scale, given that the practice rests on a small number of pioneers, such as the Dutch Tax Administration's ALEF, France's OpenFisca and the Rulemapping method used by German ministries and lacks common formats or standards.
Equipping regulators to choose and combine regulatory approaches and governance models systematically. The OECD’s report on Spectrum of Governance Models (King et al., 2026[68]) provides the analytical foundation for this: it offers a building-block method that lets governments compose models by assigning public and private influence separately across objectives, design and enforcement, a set of considerations for matching the mix to technology maturity, market structure and institutional readiness, as well as guidance on safeguards against capture, from transparency and equal-access rules to cooling-off periods and mandatory membership of self-regulatory organisations. What remains uncertain is how this emerging framework works in practice as convincing country practice in choosing, sequencing and combining models deliberately, rather than by accretion, remains thin. A second gap is institutional readiness: the more collaborative and industry-led models demand in-house expertise, digital tools and co-ordination that many regulators lack, given persistent recruitment difficulties. The OECD could develop the model-choice framework into a usable tool, test it against real regulatory decisions, and build the evidence on how governments acquire the capacity these models presuppose.
3.3.6. Public financial management systems for agility
The instruments to reconcile long-term priorities with fiscal discipline largely exist, but they are not delivering as intended. Public financial management is being asked to sustain long-term, often uncertain investment in competitiveness, the energy transition, research and development, defence and digital transformation, while holding fiscal discipline under tighter constraints than at any point since the last decade's crises. Three issues stand out, and in each the difficulty is a tension between competing goods rather than a gap to be closed through effort alone. Finding the right balance will be key for countries to move in the direction of change described in Chapter 2 and, ultimately, to restore margins of manoeuvre in their public finances.
Restoring fiscal space against path-dependency and low trust. Two main barriers stand between governments and the fiscal room they need. The first is rigidity: roughly one-third of the budget is non-discretionary across advanced economies, locked in by mandatory expenditures grounded in law, such as social entitlements and health costs, while temporary measures introduced during shocks often harden into permanent entitlements. The second is (perceived) legitimacy. Polarisation, fragmentation and low trust make it hard to frontload savings or sustain reform, and fiscal decisions are contested on perceptions of fairness as much as on substance. Poor transparency compounds this second problem: tax expenditures are growing, but governments often do not know what they cost in total. Weak fiscal ecosystems and limited public understanding of public finances weaken accountability and leave governments unable to build support for difficult trade-offs. The two challenges reinforce each other: rigidity means the savings that matter are the politically hardest to make, and low trust means those are precisely the ones governments have a hard time implementing. As a result, while savings measures are widespread, they remain modest relative to the challenges public finances are facing (see Box 3.7).
Public finances in OECD countries are facing significant challenges such as slow economic growth, ageing populations, high debt from recent crises, and new spending pressures, including in defence. About ½ of OECD countries are projected to worsen their primary deficit during 2023-2027. The 2026 OECD report Restoring Public Finances: Enabling Effective Government (OECD, 2026[40]) provides a comprehensive stocktaking of savings measures that OECD countries are implementing in 2025-2026 to make their public spending more focused, efficient and effective. Notably:
89% of OECD countries are taking savings measures in social protection and/or healthcare, which are commonly the areas with the largest spending.
77% of OECD countries are taking savings measures in government operations through efficiency-oriented efforts as well as through public employment reforms.
The report sets out key strategic directions to restore public finances, including continuing focusing on long-term sustainability in key areas of spending growth; mobilising quality budget institutions; empowering public understanding of fiscal challenges; and streamlining and simplifying processes in government operations. It serves as a knowledge base for policymakers, public servants, and civil society on the basis of the responses to the OECD 2026 Survey on Restoring Public Finances.
Source: OECD (2026[40]), Restoring Public Finances: Enabling Effective Government, OECD Publishing, Paris, https://doi.org/10.1787/fbcf9161-en.
Sustaining commitments to long-term priorities and to discipline. Long-term needs are calling for sustained public investment in competitiveness, energy infrastructure, research and development, education, environmental challenges, defence and digital transformation, at the very moment OECD countries face constrained fiscal space (OECD, 2026[40]). With limited room, funding any of these priorities means declining to fund another. Allocation is therefore a political exercise as much as a technical one, and decisions that are difficult to take prove more difficult still to sustain across time and mandates. Medium-term and top-down budgeting frameworks exist to carry fiscal prioritisation decisions and consolidation efforts across years, but they are not consistently doing so: only 7 of 26 OECD countries reported in 2023 that their most recent budget was consistent with the multi-annual expenditure ceilings set the previous year. Technical weaknesses in the design of baselines, ceilings and reserves of Medium-Term Budget frameworks play a role, as well as structural limits in the institutions meant to provide independent analysis. The biggest constraint, however, is political. A fiscal trajectory that holds only for the duration of the government that set it offers little of the predictability such frameworks are designed to provide, and sustaining one across electoral cycles depends on a degree of cross-partisan agreement that few countries have established.
Protecting investment from the path of least resistance. General government investment averaged 3.5% of GDP in 2024, down from more than 4% in 2009, even as growing and ageing populations and ageing infrastructure raise the need for it. The problem is structural: consolidation falls disproportionately on public investment precisely because it is more flexible than other spending. When countries reported their saving measures for 2025 and 2026 (see Figure 3.6), 10 declare having cut capital spending on public works and infrastructure, while only 4 chose to strengthen prioritisation or selection of capital expenditure, and 2 simplified the capital-investment framework (OECD, 2026[40]). Investment is thus often seen as the path of least resistance for short-term savings, yet much needed for long-term growth. Protecting it requires that this asymmetry be recognised in budgetary decisions rather than left implicit. Current definitions of investment offer limited assistance in this respect, as they do not readily distinguish spending that builds future capacity, particularly in human capital, research and development, and intangible assets, from spending that does not.
Measures approved or submitted to parliament for the fiscal years of 2025 and 2026
Note: Results based on 39 RPF Survey responses. Measures reported as “other” in the RPF Survey have been split into the following three subcategories, based on the qualitative information provided by respondents: “Simplify or streamline capital investment frameworks”, “Strengthen prioritisation or selection mechanisms” and “Other measures in infrastructure and public works”.
Source: OECD (2026[40]), Restoring Public Finances: Enabling Effective Government, OECD Publishing, Paris, https://doi.org/10.1787/fbcf9161-en.
These issues are not equally resolved everywhere, but a set of practices is emerging in response, each addressing one or more of the issues above.
Defining practices
Building the trust and transparency that fiscal choices depend on. Recognising that discipline depends on legitimacy, governments are strengthening the actors and information that shape how fiscal choices are understood. This means equipping Independent Fiscal Institutions, audit bodies, parliaments, civil society and media with the mandates and resources to translate complex information into accessible insight. For example, the Netherlands' Bureau for Economic Policy Analysis pairs forward looking analysis with accessible formats and a strong communication culture, and Ireland's National Economic Dialogue convenes social partners, civil society, experts and citizens before the budget to discuss priorities and trade-offs against real constraints. Legibility also depends on what is counted. Tax expenditures are among the least visible calls on public resources, and some countries are now reporting and reviewing them systematically, both to expose their cost to citizens, parliaments and government and to withdraw support that is outdated, inefficient or no longer justified. For example, in Portugal, the Ministry of Finance has established a unit responsible for evaluating and monitoring the relevance and impact of tax benefits.
Keeping mandatory spending in view and temporary measures temporary. Countries are bringing non-discretionary spending into the frame it has often escaped and making sure that temporary measures do not become permanent. Australia's Forward Estimates project all expenditure, including mandatory, three years beyond the budget. The estimate is updated and published throughout the year. The Netherlands' IBOS system system allows the Ministry of Finance and line ministries to track baseline changes in real time across all programmes, updating at any point during the year and generating standard tables that show how each programme baseline changes from year to year by reason of change, allowing emerging pressures to be identified and addressed before they accumulate into larger fiscal risks. Spending reviews increasingly examine entire policy areas including mandatory spending: Norway's 2025 review of health-related social-security benefits proposed measures to raise labour-market transitions, and a Dutch 2025 review of mental health care proposed reforms to prevention and system incentives, both presenting reform options rather than single prescriptions. On the second front, Switzerland's structural budget-balance rule permits countercyclical spending in downturns but requires offsetting surpluses in booms; the United States subjects new discretionary and mandatory legislation to Congressional Budget Office cost estimates. Similarly, New Zealand's declining real baselines, pegging most non-benefit spending at current levels with no automatic inflation adjustment, create standing incentives for efficiency and reprioritisation.
Making medium-term frameworks binding beyond the government that set them. A framework's value lies in what it can carry across mandates, and countries are strengthening the design features that let it do so. Sweden sets its aggregate ceiling in nominal terms for the budget year and three subsequent years, adding a new outermost year annually on a rolling basis; once approved by Parliament, the ceiling is normally not revised over its four-year horizon, so the anchor outlasts any single budget. Finland's four-year expenditure ceilings, renegotiated through Parliamentary cross-partisan Working Groups when funding sources change, show a different route to durability: the trajectory is agreed by a political settlement rather than to administrative rules alone.
Prioritising and redefining investment so scarce capital delivers value. Governments are sharpening how capital is selected and classified. Cost-benefit analysis, used by 84% of OECD countries surveyed in 2020 (OECD, 2020[69]), is combined with strategic-investment tests that align proposals with government objectives and appraise a wide range of interventions, physical, regulatory, demand-management and asset-management, to find the most cost-effective option. Definitions are being broadened and clarified: Canada's new Capital Budgeting Framework establishes a consistent way to classify spending that contributes to capital formation, and the UK’s stability rule includes provision for borrowing only for investment purposes, as well as an investment rule to protect capital expenditure. Some countries also relieve pressure on capital budgets by managing demand on existing assets and adopting more advanced asset management. New Zealand is one example: rather than treating the gap as a case for more capital, which would mean raising investment from 5.5% to 9.6% of GDP for three decades, it has focused on better use of existing assets, sharper project selection and non-built solutions.
Forward looking issues
The defining practices above show governments moving deliberately toward more agile public financial management. Yet several areas show promise but lack convincing country practice at scale, or raise questions no jurisdiction has fully resolved, and where the OECD is well placed to support the reflection.
Reconciling public-sector risk tolerance with the innovation long-term priorities demand. The agenda governments have set themselves, in digital, energy and competitiveness, requires experimentation and de-risking of investment, for instance through greater use of loans and blended instruments, yet the low financial-risk tolerance embedded in budgeting, audit and accountability works against it, even as major digital investments still need rigorous ex ante scrutiny to secure returns. What remains uncertain is how to accept a certain level of risk where innovation requires it without eroding the discipline that protects public money, and convincing practice is thin: the UK National Audit Office's move toward auditing "well-managed risk-taking" is an early signal rather than an established model. The OECD could document where audit and budgeting frameworks are being deliberately reoriented to sustain innovation in support of long-term growth, as well as build guidance on de-risking instruments and the assurance conditions that keep them accountable.
Turning digital tools and AI into instruments of public financial management and fiscal transparency. Real-time expenditure data and AI offer the prospect of monitoring cost pressures as they emerge and of personalising fiscal communication at scale, helping citizens see how decisions affect them directly. But practice is early and uneven: the Netherlands' IBOS shows what real-time baseline tracking can do, yet AI-enabled, citizen-facing fiscal communication remains largely aspirational, and the governance questions, accuracy, transparency and the risk of misinforming rather than informing, are unresolved. As a first step, the OECD could identify interesting practices in this rapidly evolving space.
Moving from informing citizens to giving them a real voice in fiscal choices. Strengthening the fiscal contract increasingly means involving citizens in shaping priorities, not just explaining decisions after the fact, since people are more likely to trust and support choices they feel they helped shape. Ireland's National Economic Dialogue shows the promise, but genuine deliberative practice on fiscal trade-offs remains rare, and its effectiveness hinges on a visible link between input and outcome, without which participation risks reinforcing perceptions of tokenism. The OECD could map where citizen engagement on fiscal choices moves beyond formal consultation to genuine deliberation, and identify the design conditions under which it strengthens legitimacy rather than eroding it, building on its new work on The People and The Budget (see Box 3.8) (OECD, 2026[70]).
Fiscal rules, medium-term frameworks and spending reviews remain essential to keeping public spending in check, but their effectiveness increasingly depends on securing support from citizens. The People and the Budget: Empowering Public Understanding of Public Finances (OECD, 2026[70]) argues that public understanding has become a condition of fiscal sustainability rather than a communications afterthought: reforms to taxes, pensions and public spending routinely stall not because the policies are flawed, but because citizens do not own the problem or recognise their concerns in how it is explained. With OECD public debt near 110% of GDP and projected to reach 113% by the end of 2027, and with long-term pressures from ageing, defence and extreme weather intensifying, the room for delay is narrowing.
The report sets out four pillars for Empowering Public Understanding in Public Finances:
Demystifying the budget for key decision makers: parliamentary budget offices shifting focus from budget mechanics to the strategic choices at stake, supported by well-timed inductions and case studies connecting fiscal choices to daily life.
Communicating public finances clearly: relatable comparisons and clear visuals, and a communication culture embedded within institutions.
Giving citizens a real voice: moving beyond consultation toward more ambitious forms of engagement, such as Ireland's National Economic Dialogue to.
Empowering fiscal advocates: independent fiscal institutions (IFIs) that explain long-term risks and the costs of inaction beyond expert audiences.
Drawing on the 2025 OECD Survey of Government Communicators, the 2025 OECD Survey on Parliamentary Budget Oversight, the 2024 OECD Trust Survey and the 2024 OECD Fiscal Advocacy Index, the report finds a 47% trust gap between people who feel that "people like them" have a say in government decisions and those who do not. It also finds that IFIs are most influential where strong independence is combined with focused analysis and a communications platform: publishing high-quality analysis is necessary, but rarely sufficient.
Source: OECD (2026[70]), The People and the Budget: Empowering Public Understanding of Public Finances, OECD Publishing, Paris, https://doi.org/10.1787/81674d37-en.
3.3.7. Human-centred design of public services
Key issues
Public services across OECD countries have historically been designed around ministerial mandates, legal competencies and back-office processes rather than how users actually experience them. The result is fragmented journeys, duplicated information requirements and administrative burden that fall hardest on people navigating complex life events such as losing a job or having a long-term illness. Yet, services are where most people actually meet government, which makes them the most direct route to the renewed relationship between government and citizens described in Chapter 2. It is also an essential driver of complexity for citizens and businesses experiencing government and crucial to achieving a Simple government. Three issues stand out across OECD countries.
The first concerns whether services are organised around institutional needs and processes or around people’s lives. Users experience life events – having a child, losing a job, starting a business – as single moments that require them to access multiple services, often cutting across multiple administrations and/or agencies. Where governments still organise delivery by institutional mandate rather than by life events, users have to bear the cost of navigating these different services and providing the same information to each. How people access services matters just as much. A well-designed service is of no use to the people who cannot get to it. Moving services online does not always make access easier: digitalising a service without redesigning it can make access harder for people with low digital literacy if access to in person delivery is no longer available or harder to reach. Keeping digital, telephone and in-person channels to an equivalent standard is what allows a service to reach everyone it is meant to serve. Enabling users to access services in the way they wanted is also a key driver of satisfaction (OECD-EC 2025 Survey of Citizen Satisfaction with key government services). OECD data shows that, while countries are making important progress in this direction (OECD, 2026[1]), citizens in vulnerable situations remain systematically underweighted in how services are designed.
The second concerns the gap between having standards and applying them. Whole-of-government service standards are now close to universal: 32 of the 36 OECD countries (89%) have implemented whole-of-government service standards (OECD, 2026[1]). What varies is how these are implemented and the design practice behind them. Indeed, most countries test usability with users before a service launches (81%) or run focus groups to identify what users need (69%), but far fewer use more rigorous methods to do so: just under 30% use A/B testing and around 25% use first-click or tree testing (OECD, 2026[71]). The Government service standards therefore often tend to describe what good design of public services looks like without yet changing how these are built in practice.
The third concerns whether governments actually act on information about what service users want. On average across OECD countries, only about four in ten respondents believe their government would improve a poorly performing service even if many people complained about it – a perception that has changed little since it was first measured (OECD, 2026[10]). Yet, knowledge of service performance and customer needs is increasingly widespread in government. 29 out of 31 OECD countries (93%) now collect user-experience data, by interviewing people who use public services and agency-level complaints or feedback portals are available in nearly 20 out of 31 countries (65%). However, the feedback loop back into service redesign and policy is sometimes weak: as shown in Figure 3.7, only around half of OECD countries report incorporating this information into their decision making processes to improve services (16 out of 31, 52%). As a result, tracking service performance more often functions as an accountability exercise rather than a driver of improvement.
These three issues are not equally resolved everywhere, but a set of practices is emerging in response. Five key practices stand out across OECD countries, each responding to one or more of the issues raised above.
Distribution of uses of experience and performance data in central government, 2024
Note: The figure presents responses to the question “What are the uses of the user experience survey and performance data by the central/federal government”. Responses were: “a) draft reports on service performance; b) improve specific public services; c) measure the impact of past
reforms; d) shape policy; e) report to Parliament or parliamentary commissions: f) consultations with stakeholders or the public on service
improvements; g) performance appraisals for service managers; h) performance appraisals for staff delivering services; i) communicate results
to the public; j) other”.
Source: OECD (2026[72]), Serving Citizens 2026: Delivering Better Public Administrative Services, OECD Publishing, Paris, https://doi.org/10.1787/ecaa9c7e-en.
Defining practices
These three issues are not equally resolved everywhere, but a set of practices is emerging in response. Five practices stand out across OECD countries, each responding to one or more of the issues raised above.
Moving to genuinely integrated life events. Organising services around life events spares users the burden of finding and applying to each one separately. Most OECD countries have now adopted a life-events approach in some form (22 of 30, 73%), with another 5 (17%) planning to do so [GOV/PGC(2026)16]. The life events most often covered are having a baby (11 of the 22 countries), starting a business (9 of 22) and losing a job. However, maturity across countries varies significantly: across nine common life events, around 39% of cases offer information-grouping (e.g. a single webpage collecting the relevant links) while only around 19% are genuinely integrated, with data shared across at least two agencies [GOV/PGC(2026)16]. Estonia illustrates what the integrated end of the spectrum looks like: when a child is born, the hospital submits birth data directly to the population register, which automatically cross-references tax, health, employment and social-insurance systems to determine eligibility and pay family benefits, with no application required from the parents [GOV/PGC(2026)16].
Making the once-only principle effective across every service delivery channel. Making service integration the default rather than the exception is what removes the burden of navigating government siloes from the user. No one should have to supply the same information twice to any public body, which means building the systems that share data across agencies from the outset rather than adding them later, as the Estonian example above shows in practice (see also Section 3.1 on “Trusted and connected digital technologies and data” for a discussion of these requirements). Evidence from the OECD-EC 2025 Survey on citizen-satisfaction with key government services across ten EU Member States suggests that integrated, once-only, service delivery is among the stronger drivers of user satisfaction (OECD-EC 2025 Survey of Citizen Satisfaction with key government services). Once-only design can only improve and simplify access for users if it is applied across every access channel: digital, telephone and in-person.
Using data and AI to anticipate user needs. The most advanced form of practice is a service that starts before a user knows they are entitled to it. Proactive delivery depends on the same foundations as the practices above: reliable data that can be shared across agencies, and a clear legal basis for acting on someone's behalf before they have applied. Agentic AI is beginning to extend what those foundations can do, from filling in forms in advance to routing queries to the right service automatically (see also Section 3.1 on “Trusted and connected digital technologies and data” for more on agentic AI and proactive service delivery). For example, Korea’s Benefit Notification Service (Hye-taek Allimi) uses information about an individual’s circumstances held by multiple government institutions to notify them of public services or benefits that may be relevant without requiring people to search separately. As of December 2025, this service covers approximately 6 000 services. Estonia's Unemployment Insurance Fund illustrates the more mature end of this practice: OTT2, a second-generation machine-learning tool that identifies which jobseekers are at highest risk of long-term unemployment and updates that assessment as their circumstances change, so that counsellors’ time is directed to those who need it most (OECD, 2026[73]).
Bridging the gap between service standards and design practice. Almost all OECD countries now have whole-of-government service standards, so the question is no longer what good design looks like but whether services are actually built that way in practice. Standards describe the goal services are meant to reach, whereas design methods, such as user research, journey mapping, service audits and iterative prototyping, are what get services to reach that goal. Two things help design methods take hold: giving service teams safe spaces to test, such as living labs, and rewarding the use of design methods in performance frameworks. Leading countries are moving beyond basic usability testing toward these more systematic methods, as a growing OECD case-study base of close to 800 innovative service-design practices across 83 countries shows (OECD, 2026[73]). France's Accélérateur d'Initiatives Citoyennes goes furthest, involving the public in the design itself: rather than testing a service after it is built, it gives the public a structured role in identifying where services fall short and working with public servants to redesign them [GOV/PGC(2026)16]. Japan’s experience also highlights the risk of relying on centrally designed solutions alone: while earlier centrally-designed service design reforms showed limits due to a focus on formal compliance rather than operational change, the Ministry of Internal Affairs and Communications (MIC) shifted towards building reform capability among frontline officials to equip them with do-it-yourself resources and training on service-design. Institutional hurdles identified on-site are then escalated to headquarters for review, creating a bottom-up feedback loop that informs and refines high-level policy design.
Making good design mandatory. Methods and standards only spread if someone is accountable for applying them. Most OECD countries have evolved on this: 22 of 31 (71%) countries have appointed a cross-government committee to oversee improvements to public administrative services, most with representation from several departments and senior political attendance [GOV/PGC(2026)16]. What varies is how much authority those committees hold. France's DITP illustrates a well-developed model: it leads cross-ministerial reform and acts as Secretariat to committee chaired by the Prime Minister that sets government-wide priorities, with each ministry required to set measurable targets and report progress regularly. Ireland's Public Service Leadership Board applies the same logic at working level, across departments and agencies, including police and health services. The more advanced model makes this structural rather than advisory: Significant service changes and major digital investments pass through a gateway review against user-centred criteria, such as accessibility, inclusion, once-only, before they are approved. Good design thus becomes a condition of funding rather than an aspiration.
Forward looking issues
The defining practices set out above show that some governments are moving deliberately toward more integrated, evidence-driven and proactive approaches to human-centred service design. Yet the frontier of what is achievable remains open. Several areas show genuine promise but lack convincing country practice at scale or raise questions that no jurisdiction has yet fully answered. Two stand out as priorities for further exploration, both for governments and for the OECD's future work in this space.
Making proactive service delivery the operational default. Today, for most OECD countries, proactive delivery, that is, acting on what the government already knows before a user has to ask, remains an aspiration or, at best, at pilot stage, rather than a routine feature of public service architecture. Making proactive service delivery the default would require more than just data sharing, technical interoperability and the right technology: it requires clear legal frameworks for sharing data across public institutions, accountability arrangements that allow governments to initiative a service without an explicit citizen request, and transparency mechanisms that give users meaningful visibility and control over how their data is used, as well as what steps governments take on their behalf. It also means deciding which decisions can be automated and which need human judgement. These are not questions individual service teams can answer alone. The OECD is well placed to map the legal, technical and governance preconditions for proactive delivery at scale, as well as to identify the minimum assurance conditions for responsible deployment.
Closing the loop between measurement and service improvement. Most OECD countries now collect data on how people experience public services. However, satisfaction data captures only what people perceive, so countries are beginning to pair it with direct measures of what a service costs users in time and effort. Together, the two layers tell a government not only whether people are satisfied, but whether the service is working for users and for the administration running it. The Standard Cost Model, used across OECD and EU countries to quantify the burden that regulation and administrative procedures impose, is the established method for doing. In addition to enriching existing data, what often distinguishes whether governments act on the information they hold is whether that data is comparable across institutions and published centrally. Indeed, agency-by-agency reporting tells each organisation how it is doing but does not show government as a whole which services require the most attention. This is why Estonia requires all ministries and agencies delivering public-facing services to report annually against a minimal, mandatory set of indicators – volume of services provided and user satisfaction – through standing central guidelines [GOV/PGC(2026)16]. Canada's Government of Canada Service Inventory does something similar at greater scale: a mandatory catalogue, required under the Policy on Service and Digital, covers service standards, performance data and client satisfaction data across 82 federal departments and agencies. These data are made public through the GC InfoBase dashboard [GOV/PGC(2026)16]. Building on these practices, the OECD could develop a common framework for measuring service performance that combines perception-based and objective evidence, and identify the governance conditions under which measurement drives actual service design and delivery improvement.
Together, these two areas point to what may be the defining challenge for the next phase of human-centred service reform: moving from designing better services to governing a service system that learns, adapts and reaches everyone it is meant to serve, including those least likely to ask. The OECD's continuing comparative work in this area offers a platform for surfacing the practices, frameworks and institutional conditions that will shape this next phase.
3.4.1. Introduction
As societies change, expectations rise, and policy challenges become more complex, governments can no longer rely on periodic elections or one-off consultations to understand public needs or build support for difficult reforms. Delivering the ambitions of the Government Unstuck agenda requires governments to develop continuous capabilities to listen to society, involve citizens meaningfully in shaping decisions, communicate transparently, and institutionalise these practices across government.
The OECD Trust Survey demonstrates why this matters (OECD, 2026[10]). Overall, only 41% of citizens feel they have a voice and this has consistently been overall the strongest drivers of trust in government (Figure 3.8). Yet only around 31% of people across the OECD believe that people like them have a say in what government does. While 68% believe that voting influences government decisions, confidence falls sharply for other forms of participation: only 35% believe attending townhall meetings as potentially influential, and just 32% think governments are likely to adopt the views expressed through consultations. This suggests that today, citizens increasingly expect opportunities to shape policy beyond elections but often doubt that existing engagement mechanisms make a meaningful difference. This reflects broader societal changes where citizens have larger opportunities to express their views and traditional intermediaries in policymaking have lost some of their legitimacy. The frontier between institutions and the citizens they serve is being erased at the benefit of a more co-construction approach to both policymaking and service delivery to which the institutions of representatives’ democracies have not yet fully adapted.
Share of population with high or moderately high trust in the national government by feeling they have a say in what the government does (blue) and confidence to participate in politics (yellow), 2025
Note: The figure presents the weighted country averages of responses to the question “On a scale of 0 to 10, where 0 is not at all and 10 is completely, how much do you trust the national government?” by respondents’ feeling of having a say (blue) and confidence to participate in politics (yellow). Shown here is the proportion of respondents that have “high or moderately high trust’’ based on the aggregation of responses from 6-10 on the 0-10 response scale, grouped by whether respondents feel that people like them have a say (blue): “How much would you say the political system in [COUNTRY] allows people like you to have a say in what the government does?” and feel confident to be able to participate in politics (yellow): “How confident are you in your own ability to participate in politics?” “OECD” presents the unweighted average of weighted OECD country averages.
Source: OECD (2026[10]), OECD Survey on Drivers of Trust in Public Institutions 2026 Results: Navigating Rising Expectations and New Horizons, OECD Publishing, Paris, https://doi.org/10.1787/9eb63fec-en.
Citizen engagement is therefore under pressure to evolve from isolated participation exercises into a strategic governance function. This means strengthening capabilities to understand societal concerns before they become crises, designing participation processes that genuinely inform policy choices, communicating decisions in ways that build understanding and legitimacy, and building institutional capacity to embed these practices consistently across government. When these capabilities work together, engagement becomes a continuous feedback system that improves policy quality, strengthens implementation, and reinforces democratic trust.
The section is organised around four interconnected areas of practice. Together, they describe the capabilities governments need to understand society more effectively, involve citizens more meaningfully in public decisions, build support for change in increasingly fragmented information environments, and institutionalise citizen engagement as a core function of modern government.
Defining practices for meaningful and strategic citizen engagement
3.4.2. Continuous listening and societal intelligence
Key issues
Governments have traditionally understood society through relatively infrequent and structured mechanisms such as elections, surveys, consultations and engagement with organised stakeholders and intermediaries. These channels remain essential foundations of democratic governance, but they are increasingly insufficient in a world where public attitudes, behaviours and concerns evolve rapidly, where citizens expect governments to respond more quickly to changing needs and lived experiences. Today, communication teams in most government institutions routinely analyse social media, online discussion and media discourse as major sources of continuous listening. These channels can help governments to understand emerging issues, concerns, and narratives in citizens' own terms, and use these to help shape the policy agenda and the questions on which citizens are invited to participate.
Yet these communication channels have clear limits when used on their own. Online debate is increasingly shaped by small, vocal or highly mobilised groups, while platform dynamics and opaque algorithms can distort what becomes visible. A large “silent middle” exists that is not well represented in online conversations. OECD survey evidence reinforces this point: 58% of senior communicators find that social listening offers valuable insights into how topics are perceived among some audience segments, but needs to be combined with other research to obtain an accurate picture (OECD, 2026[74]).
At the same time, governments now have access to an unprecedented range of information about society. Behavioural data, service-use data, perception surveys, digital interactions, social listening, search trends and advances in data analytics and AI create new opportunities to understand citizens' experiences in greater depth and closer to real time. Governments therefore are experimenting with broader systems that treat social listening, media monitoring and search analytics as partial inputs within a broader system of ethical societal intelligence, triangulating them with polling, qualitative research, values-based segmentation, deliberative methods, participation, and policy or service feedback. This broader mix can help distinguish signal from noise, understand the narrative environment in which citizens receive information and form views, and build a more accurate picture of public needs, expectations and priorities. Yet much of this potential remains untapped. Information is often fragmented across institutions, disconnected from policy processes, or collected only after policy priorities have already been established. The defining practices below illustrate how governments are beginning to build these capabilities through new approaches to listening, analysis, co-ordination and policy design.
Defining Practices
Strengthening continuous listening for better communication. Governments are increasingly investing in integrated capabilities to better understand evolving public concerns, information needs and media narratives. Rather than relying solely on opinion polls or ad hoc monitoring, emerging approaches combine insights from multiple information sources, including news media, social media, public feedback and communication performance data, to support more evidence-informed communication strategies. France's Government Information Service has developed an AI-powered tool to analyse trends and themes emerging across different information channels, while the European Commission has complemented data analytics with country and regional dashboards to tailor communication approaches to specific contexts and evolving trends. In Korea, the BIG KINDS platform enables systematic analysis of media coverage and emerging issues over time. Together, these initiatives leverage developments from AI to enhance listening at scale and illustrate a broader shift towards continuous monitoring, audience intelligence and data-informed decision making. They help governments better understand and even anticipate public concerns, adapt communication approaches and strengthen responsiveness. The key to their effectiveness rests in how these insights are channelled to decision makers and shape public action.
Developing richer understanding of how public views evolve. Governments are increasingly complementing traditional opinion polling with more sophisticated forms of attitudinal research that examine not only what citizens think, but how and why views change over time. These approaches help governments identify emerging concerns, understand differences across population groups, test how citizens respond to information and trade-offs, and detect shifts in trust and public sentiment before they become more deeply entrenched. Canada's Trust, Information and Digital Ecosystems Study (TIDES), for example, uses repeated nationally representative surveys to track how trust, media use, information environments and vulnerability to mis- and disinformation evolve over time, providing an early-warning system for changes in public attitudes (Government of Canada, 2024[75]). Deliberative Polling offers a complementary approach by measuring how preferences change after citizens receive balanced information and engage in structured discussion. Exercises conducted across the European Union and in Hungary found that participants' views shifted significantly on issues such as climate policy, migration, market openness and the role of government, demonstrating that initial survey responses may not fully reflect the preferences citizens form once they have had the opportunity to consider evidence and competing trade-offs (Stanford University, 2026[76]). Together, these approaches illustrate how governments can build a more nuanced understanding of public opinion, moving beyond static snapshots towards a deeper appreciation of how attitudes develop and how they may respond to different policy choices.
Forward looking issues
The defining practices above illustrate how governments are beginning to build more continuous and integrated approaches to understanding society. Yet these capabilities remain at an early stage of development, and rapid advances in AI and new forms of societal data are raising important opportunities and questions that few governments have yet addressed systematically. The following issues highlight some of the key frontiers for future policy and OECD work.
Harnessing AI for societal intelligence while safeguarding democratic values. Artificial intelligence is fundamentally expanding governments' ability to understand society by synthesising large volumes of behavioural, perception and service-use data, identifying emerging concerns and helping distinguish long-term trends from short-term noise. Yet only a small number of governments are currently experimenting with these approaches systematically. As these capabilities develop, governments and the OECD will need to address important questions around data governance, transparency and public trust, including where to draw appropriate boundaries between legitimate societal listening and inappropriate forms of citizen surveillance. The emerging use of synthetic audiences or “digital twins” for insight and testing likewise calls for better guidance and should not substitute listening to real people. Developing common principles for the ethical use of AI-enabled societal intelligence will become an increasingly important priority.
Moving from social listening to organisational listening. Looking ahead, governments will need to move beyond media monitoring and social listening towards more comprehensive forms of organisational listening. This practice requires more than analysing social and traditional media: it involves consolidating and analysing signals from multiple channels to build a deeper understanding of what matters to different groups and how policies are experienced. This shift will require stronger analytical capabilities, systems for integrating evidence across channels and points of contact between institutions and citizens, and greater capacity to translate insights into policy and communication decisions. As governments seek to become more responsive and people-centred, organisational listening is a foundation of effective public governance and communication.
3.4.3. Embedding meaningful participation across the policy cycle
Key issues
Citizen participation has expanded significantly over the past two decades as governments have introduced a wide range of participatory mechanisms at national and local levels, combining digital and face-to-face approaches to involve citizens in public decision making. Yet participation often remains concentrated at specific stages of the policy cycle after key decisions have already been made, without clear pathways to feed citizen input into decision making, and designed as standardised consultation exercises regardless of the policy challenge being addressed. Too often, these mechanisms have mixed up the consultation of citizens with that of various stakeholders, blurring the line between a democratic exercise and a policy consultation exercise. This can lead to frustration among participants and weaken links between participation and outcomes. Furthermore, traditional participation processes often attract the same highly engaged citizens and organised stakeholders, while many people face practical barriers including time, cost, accessibility, confidence or lack of awareness. The result is that participation processes are too often perceived by citizens as lacking meaningful impact on policy outcomes and as short-term public relations exercises, which can undermine trust in public institutions. Addressing this requires addressing institutional constraints as much as improving the design of individual processes. Public administrations often lack the capacities, skills and internal culture to run meaningful processes. Incentives also work against them: as pressure to deliver faster mounts, participation is often the first thing cut from policy timelines.
The goal is therefore to enable more meaningful participation by embedding it as a core governance function and a strategic input to policymaking and delivery. Different policy problems require different forms of participation, and different groups may require different pathways into participation. The voice of citizens, when properly designed to be representative, needs to be differentiated from that of stakeholders (traditional intermediaries, unions, or users of public services for example). Getting citizens’ voice calls upon the government to act in that direction, while stakeholders’ view aims at improving a policy or a service.
Well-designed and resourced participatory processes, often combining in-person and digital forms of engagement can generate structured insight on the lived experience of citizens, surface how people understand policy problems, and inform policy design grounded in everyday realities, while building the common ground needed for effective and legitimate solutions to the complex challenges governments face. To address this, governments are exploring new approaches that are explicitly designed around the nature of the decision to be made, the complexity of the issue, the degree of uncertainty involved, and the intended influence of citizen input. These approaches are supported by a healthy civic space, underpinned by freedoms of expression, association, assembly and access to information, which enables citizens to organise, contribute diverse perspectives and participate safely in democratic life.
Finally, participation only strengthens trust when it allows citizens to see how their contributions have been considered and understand the reasons behind government decisions. Recognising this, some governments are strengthening accountability and feedback mechanisms that explain how contributions have been weighed, communicate why decisions were taken, and demonstrate progress over time, while embedding these mechanisms in organisational routines that integrate citizen input throughout the policy cycle and close the feedback loop through transparent reporting and ongoing communication. The defining practices below illustrate how governments are beginning to embed participation more strategically by tailoring it to policy challenges, reducing barriers to engagement, using deliberative and collaborative approaches selectively, and strengthening accountability through visible feedback mechanisms.
Defining practices
Giving citizens a greater role in setting the policy agenda. Rather than asking citizens to comment on predefined policy proposals, governments can use participatory processes to identify the issues, concerns and priorities that should shape the agenda from the outset. Taiwan’s vTaiwan process illustrates this approach by using digital deliberation tools, including machine-learning techniques to cluster opinions and identify areas of consensus, helping governments distinguish widely shared concerns from more polarised views and informing subsequent policy discussions. Similarly, Brazil’s co-creation of its Multi-Year Plan (PPA) combined nationwide in-person consultations with the Brasil Participativo digital platform, enabling citizens and civil society organisations across the country to co-define key challenges and proposals that informed the government’s strategic planning agenda. Slovenia’s Policy Lab provides another example of institutionalised collaborative policymaking. Led by the Ministry of the Interior and Public Administration, it brings policymakers together with citizens, civil society, researchers, universities, experts and businesses from the early stages of policy development and uses various co-creation methods to develop and refine policy solutions before implementation. By enabling citizens to shape not only policy solutions but also the definition of policy challenges and priorities, these approaches can strengthen both the quality and legitimacy of public decision making.
Designing participation around the policy challenge. Governments are increasingly moving beyond standardised consultation exercises by selecting participation methods that reflect the nature of the policy challenge, the decisions to be made and the intended influence of citizen input. Lithuania's "Lithuania 2050" strategy illustrates this approach by combining large-scale participatory foresight with long-term implementation arrangements. Around 2 500 participants from across society helped co-create the country's long-term vision over a three-year period. Progress is tracked publicly through a transparent performance dashboard, demonstrating how participation can extend beyond policy design into long-term delivery and accountability. New Zealand's response to the 2019 Christchurch terrorist attack illustrates how, in contexts marked by fragile trust or social division, the design, timing and local grounding of participation can matter as much as the methods used. The government recognised that trust between affected communities and the state was fragile, and responded by conducting sustained, multi-form engagement over ten months to inform a national-level response. Feedback from over 600 stakeholders informed the vision, outcomes, and focus areas for New Zealand’s social cohesion framework. In Australia, The Victorian Auditor-General's Office developed a Public Participation in Government Decision-Making Better Practice Guide, which set out a high-level framework for agencies to use when deciding how to best involve the public in government decision making and set out principles and elements to use to audit the efficiency and effectiveness of public participation exercises.
Reducing barriers and broadening participation. Governments are increasingly recognising that broadening participation requires designing engagement around the different ways citizens participate, combining digital and face-to-face methods, working through trusted community organisations, and reducing practical barriers that prevent many people from taking part. Brazil's Brasil Participativo platform exemplifies this approach by combining nationwide digital participation with in-person state dialogues, thematic conferences and strong partnerships with civil society organisations. The model uses multiple channels to reach different communities, enabling citizens to contribute through the approaches that best suit their circumstances and helping government capture a broader diversity of perspectives. France is seeking to address participation barriers by exploring the possibility of developing a legal status for the "participating citizen", including financial compensation, dedicated citizens' leave and recognition of skills acquired through participation, helping reduce the personal and professional costs of civic engagement.
Experimenting with new participation methods and tools. Governments are increasingly using more intensive participation methods, such as citizens' assemblies, national dialogues and collaborative governance arrangements, to address policy challenges that require public learning, trust-building and sustained engagement over time. For issues involving long-term trade-offs, uncertainty or competing interests, methods such as citizens' assemblies, panels, dialogues and co-design processes can allow participants to explore evidence, deliberate collectively and develop informed recommendations. Again, here, the nature of the issue should define what is expected of these processes: in some cases, when close to a decision making goal or when there is a need to advance with citizens’ views in a debate that is stuck among stakeholders, deliberative processes with citizens only should be designed, while in other contexts, co creation among different groups informing the policy debate would be more appropriate. In some contexts, collaborative approaches can go beyond structured consultation or deliberation and extend to more sustained forms of multi-stakeholder governance. Leuven 2030 in Belgium illustrates this model by bringing together government, universities, businesses, civil society organisations and citizens in a permanent governance partnership to co-design and co-ordinate the city's transition to climate neutrality. Governments are also exploring how emerging technologies can enhance participation. For example, Nesta's Centre for Collective Intelligence Design in the United Kingdom has experimented with combining citizen input and AI to generate ideas, synthesise contributions and support collective problem-solving. These innovations point to new opportunities to expand the scale, inclusiveness and effectiveness of public participation. Colombia's National Planning Department implemented the “Tejiendo Futuros” participatory foresight methodology, where Pacific communities, youth organisations and women's organisations co-developed a Comprehensive Policy for the Development of the Pacific Region.
Strengthening accountability through visible feedback. Participation is most effective when citizens can clearly see how their contributions influence public decisions, and this influence will be more or less binding for governments depending on the process, the issue and the representativeness of the citizens engaged in the process. Governments are therefore strengthening feedback mechanisms that make participation more transparent, accountable and continuous. In Iceland, Better Reykjavik enables residents to submit, debate and prioritise proposals throughout the year, with leading ideas formally considered by the municipality and linked to participatory budgeting and policy development, demonstrating tangible policy impact over time. At the national level, Ukraine has established a government-wide system requiring ministries to publish annual consultation plans and report on participation outcomes, with results compiled centrally by the Secretariat of the Cabinet of Ministers and reviewed with civil society. These approaches demonstrate how systematic feedback and transparent reporting help strengthen the credibility, accountability and long-term effectiveness of citizen participation.
Forward looking issues
The defining practices above demonstrate how governments are making citizen participation more strategic and connected to public decision making. Yet important questions remain about how participation should evolve to respond to new technologies, changing democratic expectations and increasingly complex policy challenges. The following issues highlight key frontiers for future government practice and OECD support.
Using AI to improve and enhance meaningful participation. Artificial intelligence has the potential to transform citizen participation by helping governments engage larger and more diverse groups, lower participation barriers, analyse large volumes of citizen input and support richer deliberation at scale. For instance, government can use machine learning techniques to analyse citizens’ contributions to public consultations or surveys, surface trends, cluster opinions, and identify minority views. They can also make participation easier for all citizens by using automated translation and transcription, providing virtual assistance, and targeted communication. Finally, government can leverage generative AI and virtual reality tools to engage with citizens when redesigning public spaces or preparing for future scenarios. The OECD Typology of AI Tools to Improve and Enhance Citizen Participation can help governments navigate responsibly the landscape of existing AI tools for participation. While governments at all levels have only begun experimenting with these capabilities, and their long-term implications and the sustainability of these practices over time remain uncertain. At the same time, some countries are beginning to embed these capabilities within their participation infrastructure. For instance, Finland's Voxit and the Netherlands' Praat mee met de overheid are adapting the open-source Pol.is platform into national infrastructure that uses AI to cluster citizen views, identify areas of consensus and support large-scale online deliberation. The challenge will be ensuring that the use of AI remains a means to address existing challenges of citizen participation (e.g. accessibility, scale, resources, policy impact, design) rather than replacing human dialogue and judgement, while maintaining transparency, representativeness and public trust. Developing practical governance frameworks for AI-enabled participation should become a priority as these technologies mature while their use becomes more mainstream.
Choosing the right participation method for the right challenge. Governments now have access to an expanding range of participation methods, from digital consultations and collaborative platforms to citizens' assemblies and long-term partnerships. However, there is still relatively little guidance on how to select the most appropriate approach for different policy challenges or how to combine methods effectively across the policy cycle. Future work, such as the OECD’s forthcoming Guidebook on Strategic Citizen Participation should focus on developing practical frameworks that help governments match participation approaches to the complexity, uncertainty and political characteristics of different decisions, while embedding participation earlier in agenda-setting, policy design and implementation rather than treating it as a standalone or a tick-the-box exercise.
Reaching the disengaged and rebuilding democratic participation. Participation continues to be dominated by citizens who are already politically engaged, while those with low trust in government, limited civic skills or fewer resources remain consistently underrepresented or underheard. Yet these groups are often among those most affected by public policies and whose disengagement poses the greatest long-term risks to democratic legitimacy. Moreover, some groups are active participants in non-institutionalised forms of participation, signalling a clear willingness to express their opinions and concerns through different channels. A major challenge for governments will be developing new approaches that proactively reach beyond the "usual participants", working through trusted intermediaries, community organisations and locally embedded networks, while using digital tools responsibly to broaden participation without creating new forms of exclusion. Understanding how to engage citizens who do not normally participate may prove just as important as improving participation among those who already do.
3.4.4. Building support for change through public communication
Key issues
Governments are operating in information environments that are fundamentally different from those for which most public communication systems were designed. Citizens increasingly receive information through fragmented digital networks rather than shared public channels, and often trust information received from peers, content creators, and online communities – and curated by online information platforms algorithms - more than from public institutions. Traditional models based on broadcasting information through a limited number of media outlets are becoming less effective as audiences fragment across multiple platforms and sources.
At the same time, governments face growing pressure to deliver ambitious reforms that require public understanding, behavioural change and sustained support over time. To address this, several OECD governments are moving past the outdated conception of public communication as "selling" reforms, working instead towards using communication to create the foundations for durable public understanding, legitimacy and constructive debate. This involves communicating not only what governments are doing, but why difficult choices are necessary, how trade-offs have been weighed and how citizens' concerns have shaped decisions. This requires embedding communication and engagement early in the policy cycle to understand how different groups interpret a problem, what they see as fair or unfair, and where resistance reflects practical barriers, distrust or lived experience rather than rejection of the reform objective itself.
To achieve these objectives some governments are adapting communication to the plain language, formats, tone and storytelling styles that make digital creators effective and trusted, and are partnering with creators, community leaders, civil society organisations, frontline professionals, journalists and other intermediaries who often have the platform literacy, audience knowledge and perceived authenticity that institutions lack, particularly in online spaces shaped by peer-to-peer communities (Figure 3.9). This is especially important when the aim is to reach different opinion segments, including people who are sceptical of government or disengaged from mainstream channels.
Has your institution formally collaborated with any of the following in the last year as part of its communication campaigns and activities?
Note: Data shown above refers to 56 responses to the relevant question in the OECD Survey of Government Communicators. Multiple choices possible.
Source: OECD (2026[74]), Reforming Government Communication for the Age of Algorithms, OECD Publishing, Paris, https://doi.org/10.1787/adef6a50-en.
Modern public communication also requires ringfencing its trustworthiness as a public-interest function, rather than a vehicle for political messaging. In polarised settings, communication that is fronted by politicians or seen as supporting a political argument can undermine institutional credibility and weaken efforts to build trust with those who hold different political views. Even if political and institutional communication functions are distinct in most countries, 65% of surveyed public communicators report increased politicisation of their function as detrimental to efforts for building public trust.
The defining practices below illustrate how governments are beginning to modernise public communication by adapting to fragmented information environments, working through trusted intermediaries, building public understanding around difficult reforms, and strengthening information integrity.
Defining practices
Communicating through trusted voices and relatable content. Governments are increasingly redesigning public communication to reflect how citizens consume information today rather than relying on traditional broadcast approaches. The United Kingdom's New Media Unit illustrates this shift by producing highly localised and explicitly apolitical communication about what changes on the ground for citizens. These are designed for the digital platforms citizens use every day. Rather than following the political news cycle, the Unit focuses on demonstrating tangible improvements in people's lives through place-based storytelling and trusted messengers such as frontline workers, community figures and content creators. By combining evidence on audience behaviour with platform-specific content and continuous evaluation, the approach has achieved stronger engagement, greater understanding and more positive perceptions of government delivery, and is now being scaled across government. Similarly, the European Commission's localisation strategy focuses on localising its communication to make it more relevant to the communities it serves. It works through Representation Offices, local partners and "local heroes" to adapt European policy messages to regional contexts and communicate through local media, multilingual content and community-based storytelling. By connecting policy messages to citizens' everyday experiences and allowing trusted local actors to communicate them, the approach has increased the visibility and relevance of EU policies across diverse communities.
Embedding communication throughout the policy cycle. Governments are increasingly repositioning communication as a core component of policymaking rather than a function focused primarily on announcing or explaining decisions after they have been made. The United Kingdom's Government Communication Service Toolkit: Communication for Citizen Participation illustrates this shift by positioning communication teams as providers of public insight throughout the policy cycle. The toolkit encourages communicators to analyse audience and stakeholder perspectives, identify who should be informed and involved, support participation processes, and communicate outcomes back to citizens, ensuring that societal intelligence informs agenda-setting and policy design from the outset. Some governments are reinforcing this approach through multidisciplinary teams that bring together policy, communication, engagement and delivery expertise around complex challenges. In Ireland, for example, a 2023 task force on environmental policy included communication specialists alongside policy and other experts, while several UK departments have embedded communication professionals directly within flagship policy teams. Together, these approaches demonstrate how integrating communication expertise into policy development can strengthen public understanding, improve policy design and create more effective feedback loops between government and society.
Forward looking issues
The defining practices above demonstrate how governments are adapting public communication to a rapidly changing information environment. Yet the pace of technological change and evolving expectations of public communication are creating new challenges that few governments have yet addressed systematically. The following issues highlight key frontiers for future government practice and OECD support.
Navigating AI-generated information environments. Generative and agentic AI are transforming how citizens discover, consume and create information. Governments will need to rethink public communication for environments where personalised AI assistants, synthetic content and conversational interfaces increasingly mediate access to information. This raises new questions about authenticity, transparency, institutional visibility and public trust. AI is of course also opening up vast new possibilities for communication, particularly to overcome the constraints of producing and tailoring vast amounts of content to better connect with different segments of the population.
Building public understanding around difficult reforms. Governments are increasingly recognising that communication should not simply promote reforms but help citizens understand and deliberate on difficult choices, the trade-offs involved and how different perspectives have influenced final decisions. The OECD’s guidance on Public Communication Handbook to Bridge Divides and Shape Better Policy: Voices that Resonate explores how integrating advanced communication practices, listening and engagement throughout the policy cycle can create and sustain a constructive space for policy debate on contested policy issues where most citizens can feel heard and engage. This space forms the foundation to arrive at policies that are more responsive to citizens’ concerns and more likely to command broad and lasting support. The guidance focuses in particular on how narratives act as filters to make sense of complex issues and interpret new information about them, and explores how institutions can facilitate the uptake of narratives that are more conducive to exchange than conflict. The guidance additionally stresses communication as part of the groundwork that goes into early stages of policy development and paves the way for more socially acceptable reforms.
3.4.5. Building institutional capability for participation at scale
Key issues
Most governments have implemented innovative participation practices, deliberative processes, digital engagement tools and communication approaches. However, these efforts often remain fragmented, unevenly distributed across government and heavily dependent on individual champions, temporary funding or isolated initiatives. Many governments have dedicated participation centres, cross-government practitioner networks and common standards, but still struggle to transform participation from a series of disconnected projects into a coherent government capability, reducing duplication, improving quality and enabling continuous learning across institutions. The challenge is therefore increasingly institutional rather than methodological. Moving beyond isolated engagement exercises towards permanent institutional capability will require establishing clear leadership and co-ordination arrangements, developing professional skills across the public sector, embedding participation expectations into policy processes, creating robust accountability and evaluation frameworks, and ensuring that participation becomes part of routine policymaking rather than an exceptional activity.
This transformation also depends on strengthening the capabilities of the public service itself and developing a culture of participation within the public sector. In leading countries, participation is increasingly regarded as a core competency of policymaking rather than a specialist skill possessed by a small number of engagement professionals. Public servants need the skills to design, facilitate and evaluate participation processes, work collaboratively with citizens, civil society, and elected politicians, and make effective use of behavioural insights, communication and emerging digital technologies. At the same time, governments can develop broader professional communities that connect officials with practitioners, researchers and civil society organisations, recognising that participation expertise is developed through collaboration and shared practice as much as through formal training.
Finally, institutionalising participation requires stronger accountability and a modern digital infrastructure. Governments are looking to develop better ways to assess both individual participation processes and the overall health of their participation ecosystem. Beyond evaluating whether a particular exercise was successful, governments increasingly need a system-wide understanding of participation activities, capabilities, quality and outcomes across institutions. This can support strategic decision making, continuous improvement and more effective investment in participation capacities. In addition, many of the most innovative participation practices continue to emerge at local level, highlighting the importance of creating mechanisms that spread successful approaches across ministries and levels of government. At the same time, governments are increasingly recognising the need to monitor civic space more systematically in order to gain a clearer understanding of emerging trends, areas of concern, and opportunities for improvement.
The defining practices below illustrate how governments are strengthening the institutional foundations needed to make citizen participation a routine capability of public administration rather than a collection of isolated initiatives.
Defining practices
Strengthening institutional leadership and co-ordination to position participation as a core government function. Governments are increasingly establishing dedicated institutions to co-ordinate participation across government, helping to move engagement from isolated projects to a consistent feature of policymaking. France's Interministerial Centre for Citizen Participation (CIPC) provides advice, training and methodological support to all ministries while co-ordinating a cross-government network of participation practitioners. Between 2022 and 2023, it supported 17 of the 22 national participation processes launched by government, involving more than 400 000 citizens, and evaluations show that participation processes supported by the Centre achieved a 60% improvement in methodological quality. Similarly, New Zealand's Policy Project, housed within the Department of the Prime Minister and Cabinet, provides common guidance, practical tools, communities of practice and cross-government learning to strengthen participation capability across the public service. Together these approaches demonstrate how central leadership can improve quality, build institutional memory and create economies of scale.
Building participation capability across the public service. Governments are increasingly embedding participation as a core competency of policymaking rather than treating it as a specialist discipline. This requires equipping public servants with the skills to design, facilitate and evaluate participation processes, alongside the institutional incentives to use them consistently. New Zealand's Policy Skills Framework incorporates citizen engagement directly into recruitment, performance management and capability development for policy officials, making participation an expected element of quality policy advice. Australia's Office of Impact Analysis reinforces this approach by embedding consultation as one of the seven impact analysis questions of the Australian Government's Impact Analysis Framework to explain who was consulted, how participation was undertaken, key feedback including areas of agreement and difference and where citizen feedback influenced policy options, supported by practical guidance on effective consultation. Together, these approaches help make participation a routine expectation of policymaking rather than the responsibility of specialist engagement teams.
Strengthening accountability and measurement. Governments are increasingly recognising that participation should be assessed by the quality, representativeness and influence of engagement on public decisions. The Victorian (Australia) Auditor-General's Office (VAGO) has developed one of the most comprehensive approaches through its Better Practice Guide and systematic performance audits of public participation. Rather than measuring activity alone, VAGO assesses whether participation processes meet recognised standards and examines how effectively citizen input has informed government decision making. Its audits have also created an evidence base for continuous improvement, demonstrating how independent evaluation can strengthen both accountability and the long-term quality of participation practices. Governments are also increasingly advancing towards better data collection and disclosure of relevant data for civic space protection and promotion. For instance, as part of Finland’s “CSO Strategy 2023-2027”, the country published its first report on “Basic information about the operating conditions of CSOs”, providing an overview of the CSO sector that is helping to inform policymaking for the sector. The country also launched a ready-made visualisation tool gathering existing data from different public sector sources.
Connecting participation across levels of government. Many of the most innovative participation practices originate at local level but struggle to influence national policymaking. Governments are therefore creating institutional mechanisms to connect participation across levels of government and accelerate the spread of successful approaches. Spain's Open Government Forum provides one such model by bringing together central government, regional governments, local authorities and civil society to co-create and monitor national Open Government Action Plans. Through the participation of the Spanish Federation of Municipalities and Provinces and its network of more than 200 municipalities, local participation experience is systematically captured and fed into national policy development, demonstrating how governments can transform isolated local innovation into a shared national capability.
Forward looking issues
Building the institutional infrastructure and co-ordination mechanisms for participation. While many governments have demonstrated successful participation initiatives, more can be done to embed participation systematically into the core operating systems of government. The next frontier is moving beyond isolated projects towards durable institutional arrangements that make citizen participation a routine input into agenda-setting, policy development, implementation and evaluation. This will require participation to be reflected in governance frameworks, policy development processes, accountability systems and digital infrastructure, supported by clear mandates, stable resourcing and co-ordination mechanisms across ministries and levels of government, rather than remaining dependent on individual champions or ad hoc funding.
Developing the skills, tools and culture for participation across the public sector. Institutional arrangements only deliver if public servants have the capability and motivation to use them. Yet participation skills remain unevenly distributed, concentrated in specialist teams rather than treated as a core policy competency, and rarely reflected in recruitment, training or career progression. Incentive structures often work against participation: under delivery pressure, engagement is frequently the first element cut from policy timelines, and officials are seldom rewarded for engaging citizens well. The frontier is building a genuine culture of participation, i.e., recognising engagement as a core policy skill and integrating it into leadership expectations and performance frameworks, rather than treating it as a compliance exercise.
Developing institutional learning through more accountable and impactful participation ecosystems. Governments are increasingly recognising that evaluating individual participation processes is not sufficient to build participation at scale. Emerging approaches seek to create government-wide systems that provide a comprehensive understanding of how participation functions across public institutions. These systems monitor where and how participation is being used, which communities are being reached, the quality and representativeness of participation processes, the policy issues being addressed, and the capabilities available across ministries and agencies. Moving beyond compliance and reporting, they aim to generate institutional learning by aggregating evidence across multiple participation initiatives, identifying recurring challenges and successful practices, and supporting continuous improvement. By treating participation as a government capability that can be monitored, benchmarked and strengthened over time, these approaches help governments move from isolated evaluations towards a more strategic understanding of the health, effectiveness and maturity of their participation ecosystem.
Scaling and embedding digital infrastructures for participation. Governments are increasingly using civic technologies and artificial intelligence to broaden participation, collect and analyse large volumes of citizen input, and enable more continuous engagement. However, the use of civic tech tools is often discontinued or weakened by institutional and political changes, which limits the potential of these tools to become established, channel meaningful participation, and lead to policy impact. Some countries are already embedding civic tech tools in their participation infrastructure while also connecting them with the broader digital government ecosystem. For instance, Brazil developed the digital participation platform Brasil Participativo, accessible through the Brazilian digital identity system Gov.Br. The federal unit responsible for the development and maintenance of the platform, the Secretariat for Social Participation, is also providing technical support to State-level and local authorities to deploy their own participation platforms. Moving forward, governments should shift from the ad hoc adoption of digital tools for participation to the creation of digital infrastructures for participation embedded in the broader digital government ecosystem.
Moving towards a whole-of-government approach to civic space monitoring. Looking ahead, advancing the monitoring of civic space will require governments to move beyond ad hoc data collection towards more integrated and participatory monitoring systems. This will include strengthening the use of administrative data, improving interoperability and data-sharing across public institutions, and developing common indicators that allow for the systematic tracking of trends over time. Governments could also benefit from establishing closer partnerships with CSOs and independent oversight bodies to complement official data with qualitative insights and broader evidence. Increased transparency through the publication of monitoring data and user-friendly visualisation tools can further support public accountability and informed policymaking. Embedding civic space monitoring within broader governance and public sector reform efforts can help ensure a whole-of-government approach, enabling governments to identify emerging challenges more effectively and assess the impact of policy on civic space.
References
[36] Australian Public Service Commission (2021), The APS Mobility Framework, https://www.apsc.gov.au/initiatives-and-programs/aps-mobility-framework.
[24] Danish Agency for Digital Governance (2025), Danish Strategies for artifical intelligence, https://en.digst.dk/digital-governance/new-technologies/danish-strategies-for-artificial-intelligence/.
[23] DITP (2024), le Campus de la transformation, Direction interministerielle de la transformation publique, https://www.modernisation.gouv.fr/accompagner-les-administrations/campus-de-la-transformation.
[8] DNP (2025), CONPES 4144: La hoja de ruta de Colombia en Inteligencia Artificial para los retos actuales y la transformación futura, Departamento Nacional de Planeación, https://www.dnp.gov.co/publicaciones/Planeacion/Paginas/conpes-4144-hoja-de-ruta-colombia-inteligencia-artificial-retos-actuales-transformacion-futura.aspx.
[51] Draghi, M. (2024), The future of European competitiveness: In-depth analysis and recommendations, Publications Office of the European Union, https://commission.europa.eu/document/download/ec1409c1-d4b4-4882-8bdd-3519f86bbb92_en?filename=The%20future%20of%20European%20competitiveness_%20In-depth%20analysis%20and%20recommendations_0.pdf.
[5] European Commission (2026), European Digital Identity, https://commission.europa.eu/topics/digital-economy-and-society/european-digital-identity_en.
[4] European Union (2018), “The single digital gateway”, Summaries of EU Legislation, Last update 21.11.2022, https://eur-lex.europa.eu/EN/legal-content/summary/the-single-digital-gateway.html.
[52] Federal Government of Mexico (2025), Plan Mexico: Estrategia de Desarrollo Económico Equitativo y Sustentable para la Prosperidad Compartida, https://www.planmexico.gob.mx/.
[21] Finland Ministry of Education and Culture (2026), Results of two-year pre-primary education trial published, https://valtioneuvosto.fi/en/-/1410845/results-of-two-year-pre-primary-education-trial-published.
[9] Gartner (2025), Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027, https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027.
[25] GOV UK (2023), National Digital Twin Programme (NDTP), https://www.gov.uk/government/collections/the-national-digital-twin-programme-ndtp.
[11] Government of Austria (2015), Federal Act amending the Family Burdens Equalisation Act 1967 (Federal Law Gazette I No. 50/2015), https://www.ris.bka.gv.at/Dokumente/BgblAuth/BGBLA_2015_I_50/BGBLA_2015_I_50.pdf.
[34] Government of Canada (2025), Public Service Commission of Canada’s 2025-2026 Departmental Plan, https://www.canada.ca/en/public-service-commission/services/publications/2025-2026-departmental-plan.html.
[75] Government of Canada (2024), Trust, Information, and Digital Ecosystems Study (TIDES).
[46] Government of Finland (2023), Ministerial Committees and Working Groups, https://valtioneuvosto.fi/en/government/ministerial.
[22] Government of Flanders (2024), Strategic Foresight in an international context, https://www.vlaanderen.be/en/authorities/foreign-policy/flanders-and-the-eu/strategic-foresight-in-an-international-context.
[53] Government of France (2025), Schéma de promotion des achats publics socialement et écologiquement responsables 2025-2027, https://www.economie.gouv.fr/dae/clauses-sociales-dans-les-marches-publics-un-impact-concret-et-durable-sur-lemploi.
[54] Government of France (2025), The rules of the doctrine, https://www.numerique.gouv.fr/offre-accompagnement/cloud-administrations/programme.
[37] Government of Ireland (2025), People Standards Framework, https://www.gov.ie/en/department-of-public-expenditure-infrastructure-public-service-reform-and-digitalisation/campaigns/people-standards-framework/.
[32] Government of Singapore (2024), Unique Growth Opportunities Public Officers Can Take Advantage Of, https://www.psd.gov.sg/challenge/articles/unique-growth-opportunities-public-officers-can-take-advantage-of/.
[31] Government of the United Kingdom (2025), AI Skills Competency Framework, https://www.gov.uk/government/publications/ai-skills-for-the-uk-workforce/ai-skills-tools-package#the-ai-skills-framework-mapping-ai-skills-by-job-level.
[68] King, B. et al. (2026), “The spectrum of regulatory models and its application to the digital economy”, OECD Regulatory Policy Working Papers, No. 26, OECD Publishing, Paris, https://doi.org/10.1787/b7456763-en.
[18] NZ DPMC (2026), Government Targets.
[20] OECD (2026), Anti-Corruption and Integrity Outlook 2026: Harnessing the Integrity Advantage, OECD Publishing, Paris, https://doi.org/10.1787/16708b78-en.
[7] OECD (2026), Building an AI-ready public workforce: Implications and strategies, OECD Publishing, Paris, https://doi.org/10.1787/b89244c7-en.
[43] OECD (2026), Building Centre of Government Capabilities to Steer and Deliver Complex Priorities: Synthesis Report, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/9899e2fe-en.
[35] OECD (2026), Developing a Resilient HR Function in the French Public Service: Addressing Digital, Ecological and Managerial Change, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/a6bf832b-en.
[71] OECD (2026), “Digital Government Index and Open, Useful and Re-usable Data Index: 2025 Results and Key Findings”, OECD Working Papers on Public Governance, Vol. No. 90, https://doi.org/10.1787/6347ec74-en.
[1] OECD (2026), Digital Government Outlook 2026: From Foundations to Transformational Impact, OECD Publishing, Paris, https://doi.org/10.1787/0496b2bc-en.
[39] OECD (2026), Forging New Frontiers in Mission-Oriented Innovation Policies, OECD Publishing, Paris, https://doi.org/10.1787/d13d0142-en.
[73] OECD (2026), Global Trends in Government Innovation 2026: Government Unstuck, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/0162a8ce-en.
[14] OECD (2026), Government Unstuck Survey.
[29] OECD (2026), NSG Survey on AI in the Public Service.
[10] OECD (2026), OECD Survey on Drivers of Trust in Public Institutions 2026 Results: Navigating Rising Expectations and New Horizons, OECD Publishing, Paris, https://doi.org/10.1787/9eb63fec-en.
[59] OECD (2026), “Public procurement in Costa Rica: Manual to strengthen market engagement and advance technological neutrality and competition”, OECD Public Governance Policy Papers, No. 84, OECD Publishing, Paris, https://doi.org/10.1787/d94a03ac-en.
[61] OECD (2026), Public Procurement, Trade and Industrial Policies: Supporting National Priorities, OECD Publishing, Paris, https://doi.org/10.1787/64f22ae6-en.
[74] OECD (2026), Reforming Government Communication for the Age of Algorithms, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/adef6a50-en.
[40] OECD (2026), Restoring Public Finances: Enabling Effective Government, OECD Publishing, Paris, https://doi.org/10.1787/fbcf9161-en.
[72] OECD (2026), Serving Citizens 2026: Delivering Better Public Administrative Services, OECD Publishing, Paris, https://doi.org/10.1787/ecaa9c7e-en.
[62] OECD (2026), Smart Regulations, Strong Business: Enabling Growth and Societal Protection in a Fast‑Changing World, OECD Publishing, Paris, https://doi.org/10.1787/93d38770-en.
[70] OECD (2026), The People and the Budget: Empowering Public Understanding of Public Finances, OECD Publishing, Paris, https://doi.org/10.1787/81674d37-en.
[67] OECD (2025), Best practice principles for licensing and permitting: How to manage risks while supporting growth, OECD Publishing, Paris, https://doi.org/10.1787/5f63586d-en.
[2] OECD (2025), Digital Government in Australia: Enhancing Digital Investment, OECD Digital Government Studies, OECD Publishing, Paris, https://doi.org/10.1787/91c22326-en.
[56] OECD (2025), Enhancing Public Procurement through Collaboration in Slovenia: The Way Forward, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/2eeb3463-en.
[6] OECD (2025), Governing with Artificial Intelligence: The State of Play and Way Forward in Core Government Functions, OECD Publishing, Paris, https://doi.org/10.1787/795de142-en.
[47] OECD (2025), Government at a Glance 2025, OECD Publishing, Paris, https://doi.org/10.1787/0efd0bcd-en.
[50] OECD (2025), Implementing the OECD Recommendation on Public Procurement in OECD and Partner Countries: 2020-2024 Report, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/02a46a58-en.
[63] OECD (2025), OECD Regulatory Policy Outlook 2025, OECD Publishing, Paris, https://doi.org/10.1787/56b60e39-en.
[58] OECD (2025), Promoting Strategic and Green Public Procurement in France: Professionalising the State Procurement Function, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/70da2048-en.
[15] OECD (2025), Workforce Insights from Central Governments: Findings of the 2024 OECD/EU Survey of Public Servants, OECD Publishing, Paris, https://doi.org/10.1787/2f9080b1-en.
[57] OECD (2024), Improving Competitive Practices in Hungary’s Public Procurement: Reducing Single-bids and Enhancing Supplier Participation, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/5d1c1ec1-en.
[27] OECD (2024), LOGIC: Good Practice Principles for Mainstreaming Behavioural Public Policy, OECD Publishing, Paris, https://doi.org/10.1787/6cb52de2-en.
[49] OECD (2024), OECD Public Procurement Survey.
[66] OECD (2024), OECD Survey on Drivers of Trust in Public Institutions – 2024 Results: Building Trust in a Complex Policy Environment, OECD Publishing, Paris, https://doi.org/10.1787/9a20554b-en.
[12] OECD (2024), “Recommendation of the Council on Artificial Intelligence”, OECD Legal Instruments, OECD/LEGAL/0449, OECD, Paris, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449.
[44] OECD (2024), “Recommendation of the Council on the Governance of Critical Risks”, OECD Legal Instruments, OECD/LEGAL/0405, OECD, Paris, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0405.
[60] OECD (2024), Strengthening Oversight of the Court of Auditors for Effective Public Procurement in Portugal: Digital Transformation and Data-driven Risk Assessments, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/35aeab1e-en.
[55] OECD (2023), OECD Review of Gender Equality in Colombia, OECD Publishing, Paris, https://doi.org/10.1787/a559fc5e-en.
[28] OECD (2023), Public Employment and Management 2023: Towards a More Flexible Public Service, OECD Publishing, Paris, https://doi.org/10.1787/5b378e11-en.
[16] OECD (2023), Strengthening Policy Development in the Public Sector in Ireland, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/6724d155-en.
[65] OECD (2022), Equipping Agile and Autonomous Regulators, The Governance of Regulators, OECD Publishing, Paris, https://doi.org/10.1787/7dcb34c8-en.
[13] OECD (2021), “Recommendation of the Council for Agile Regulatory Governance to Harness Innovation”, OECD Legal Instruments, OECD/LEGAL/0464, OECD, Paris, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0464.
[69] OECD (2020), Value for Money, OECD Infrastructure Toolkit, OECD, Paris, https://infrastructure-toolkit.oecd.org/governance/value-for-money/ (accessed on 13 July 2026).
[48] OECD (2015), “Recommendation of the Council on Public Procurement”, OECD Legal Instruments, OECD/LEGAL/0411, OECD, Paris, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0411.
[64] OECD (forthcoming), Accelerating Infrastructure Permitting, OECD Publishing, Paris.
[33] OECD (forthcoming), Public Employment and Management 2026.
[42] OECD/European Commission (2025), Measuring Citizen’s Satisfaction with Key Government Services for Better Performance and Enhanced Trust.
[26] OPSI (2025), “Bringing governments together to tackle common innovation challenges”, OECD Gov2Gov innovation incubator, Observatory of Public Sector Innovation, https://oecd-opsi.org/work-areas/gov2gov-innovation-incubator/.
[30] Republic of Latvia (2026), The first Artificial Intelligence Day for Public Administration concluded with practical workshops and discussions.
[76] Stanford University (2026), What is Deliberative Polling?.
[3] State Chancellery of Schleswig-Holstein (2024), Open Innovation and Open Source Strategy of Land Schleswig-Holstein, https://www.schleswig-holstein.de/DE/landesregierung/themen/digitalisierung/linux-plus1/Service/Downloads/_dateien/open-source-strategy_EN.pdf?__blob=publicationFile&v=3.
[45] Tõnurist, P. and J. Orlik (2025), “Towards anticipatory governance guidelines for public sector organisations”, OECD Working Papers on Public Governance, No. 82, OECD Publishing, Paris, https://doi.org/10.1787/a5203d0b-en.
[19] UAE (2024), Zero Government Bureaucracy Programme.
[41] UK Cabinet Office (2022), Guidance on the undertaking of Reviews of Public Bodies, https://www.gov.uk/government/publications/public-bodies-review-programme/guidance-on-the-undertaking-of-reviews-of-public-bodies.
[17] UK Civil Service (2026), Performance management framework for the Senior Civil Service (2025 to 2026 performance year).
[38] United Kingdom Government People Group (n.d.), Civil Service People Plan 2024-2027 (HTML), https://www.gov.uk/government/publications/civil-service-people-plan-2024-2027/civil-service-people-plan-2024-2027-html#high-performing-hr-function.