Why Nothing Gets Done
What Britain's policy institutions have found about government delivery — and where they agree.
Executive Summary
This paper synthesises 101 publications produced between 2020 and March 2026 by eight of Britain's most significant policy institutions: the Institute for Government (IFG), the Tony Blair Institute for Global Change (TBI), Policy Exchange, the National Audit Office (NAO), the Institute for Public Policy Research (IPPR), the Centre for Policy Studies (CPS), Reform, and the Centre for British Progress (CBP). These organisations span the full width of the British political spectrum. They disagree on tax, on the size of the state, on immigration, on the role of markets. What they don't disagree on is this: the British government cannot reliably deliver what it promises.
That consensus, documented here in granular detail for the first time, is the central finding of this paper.
The diagnosis is settled. Every serious policy institution in Britain has now produced substantial evidence that the UK state fails systemically at implementation. The IFG's annual Performance Tracker shows eight of nine public services performing worse in 2024 than in 2010, despite spending rising across the board. The NAO's audit data shows only around 11% of major government projects are rated green by the Infrastructure and Projects Authority, meaning on track to succeed. TBI's analysis of public sector productivity shows inputs to public services rose nearly 25% between 2019 and 2024, while output grew only 14%. The CPS finds public sector productivity still 5.7% below 2019 levels. Reform's workforce surveys document a civil service in which performance management is weak, specialist skills are scarce, and efficiency incentives are structurally absent. This is not contested terrain. It's a shared empirical reality.
Six areas of strong consensus emerge from the literature.
First, the centre of government is too weak to drive delivery. No.10, the Cabinet Office, and the Treasury are structurally misaligned institutions that have failed successive prime ministers. The IFG's Commission on the Centre of Government spent a year gathering evidence from former prime ministers and permanent secretaries and reached this conclusion unambiguously. TBI and IPPR reach the same finding from different angles. So does Policy Exchange.
Second, civil service capability has eroded. The data here is stark and cross-partisan. Over half of top civil servants attended Oxbridge. Under a quarter have STEM degrees. Grade inflation has exploded — Grades 6 and 7 rose 121% since 2010. Specialist skills in digital, commercial, and project management are chronically short. Reform, IFG, Policy Exchange, and TBI all document this in detail.
Third, major project delivery is structurally broken. Only around 11% of government major projects are rated green by the IPA. The Government Major Projects Portfolio carries a whole-life cost of £834 billion as of March 2024. The NAO's megaprojects analysis, IFG's data, and CBP's infrastructure work all conclude that the UK's record by international standards is catastrophically bad.
Fourth, digital and AI transformation is necessary but under-resourced. TBI estimates £40 billion a year in potential productivity gains from AI. The NAO has traced persistent structural barriers to digital delivery — wrong sequencing, legacy system costs, supplier dependency, absent revenue funding. IFG cautions that the technology itself isn't the hard part. The agreement on the prize coexists with disagreement on the delivery vehicle.
Fifth, accountability mechanisms have become obstacles to delivery. Public inquiries take years and cost hundreds of millions. Judicial review creates strategic delay. Procurement rules generate compliance theatre. Planning and environmental regulation blocks infrastructure for decades. This is the politically charged consensus — it cuts across left and right, but the institutions agree: the UK's accountability architecture produces more process than accountability, and more delay than scrutiny.
Sixth, the Treasury controls too much and delivers too little. Cross-party frustration with Treasury dominance over delivery departments runs through the IFG, TBI, and IPPR. Departmental bilateral spending deals reinforce silos. Growth policy lacks strategic capacity at the centre. The 2025 Spending Review, by multiple accounts, failed to use missions to overcome this.
Three areas of genuine dispute remain.
Should the state get bigger or smaller? IPPR and TBI want more capacity, more institutions, more funded reform. CPS and Policy Exchange want a smaller state doing fewer things better. This is a real philosophical divide.
Should delivery be centralised or devolved? TBI and IFG broadly favour strengthening the centre with mission boards and delivery units. CBP wants powerful development corporations that bypass central government entirely. IPPR wants more powers for combined authorities. Policy Exchange is split.
Is technology the solution or the supplement? TBI is the most optimistic: £40 billion a year, reimagined services, a state that barely resembles today's. IFG is cautious: the technology isn't the hard part, the institutions are. NAO documents a decade of digital failure. Reform sits between the two.
Blueprint's assessment is that the argument about diagnosis has been won. What the institutions can't protect is the remedy. Every serious reform — delivery units, spending frameworks, capability programmes, mission boards — has been reversed, diluted, or simply not implemented when governments changed. The obstacle to better delivery isn't ideas. It's the structural vulnerability of reform to political churn. The case for statutory, independently monitored delivery architecture — something with the permanence of the Climate Change Act — has never been stronger, and has never been made.
Part 1: The Landscape
Who's Working on This
Britain has never had more organisations studying why it can't deliver. The eight institutions covered in this synthesis have between them produced over 100 significant publications on government delivery in the six years from 2020 to March 2026. That's not a sign of a solved problem. It's a sign of a recognised crisis.
The institutions cluster into four rough groups by analytical character. The IFG and NAO are primarily evidence-producing bodies — they track data, audit performance, and document failure. TBI and IPPR are primarily prescription-generating bodies — they diagnose problems and propose solutions, with a broadly progressive instinct. Policy Exchange, CPS, and Reform are primarily reform-advocacy bodies — they argue for structural change, with a broadly conservative instinct. CBP occupies its own niche: infrastructure-focused, technocratic, and impatient with both left and right process instincts.
None of these categories is rigid. The IFG produces detailed proposals. Reform publishes empirical data. TBI generates evidence as well as prescription. But understanding the institutional character of each body matters for reading what follows.
The Institute for Government
The IFG is the largest single contributor to the delivery literature in this synthesis. Its 44 publications cover every major dimension of delivery failure: annual public services data through the Performance Tracker; annual civil service data through Whitehall Monitor; structural analysis of the centre of government through its year-long Commission; major project data; digital government assessments; procurement accountability; and local government delivery.
The IFG's authority derives from its consistency. It doesn't shift position with governments. It told the Blair government that public service improvement was slower than spending suggested. It told the Coalition that headcount cuts without workforce planning would damage capability. It told Johnson that his delivery unit lacked the conditions for success. It told Starmer that mission boards were inadequate for mission delivery. The IFG's analysis isn't partisan — it's relentless. And what it relentlessly finds is that the state can't deliver.
The trend in IFG output since 2020 has been an escalating tone of urgency. The 2020 Performance Tracker documented fragility. The 2023 tracker described services that were "crumbling." The 2026 Whitehall Monitor concluded that Labour had "missed" its make-or-break year on reform. The IFG isn't crying wolf. It's documenting a structural deterioration over a decade and a half.
The Tony Blair Institute
TBI has become the most prolific source of delivery prescription in Britain, particularly since July 2024. Its 28 publications are organised around a single thesis: the UK government's existing operating model is structurally incapable of meeting citizens' expectations, and AI offers the only viable path to a qualitatively different kind of state. TBI calls this the "Reimagined State."
TBI's influence on current government thinking is substantial. The mission boards concept, the delivery unit structure, the digital ID agenda, the AI Opportunities Action Plan — all carry TBI fingerprints. The institute's May 2024 flagship paper, Governing in the Age of AI, proposed an "AI Mission Control" at Number 10, a "Bezos mandate" requiring departmental APIs, and a Digital Public Assistant for every citizen by 2030. The £40 billion per year productivity estimate from that paper has become a reference point across the policy debate.
TBI's output is growing, not plateauing. It produced more publications in 2025 alone than in any previous year, reflecting both the institute's proximity to the current government and its conviction that the window for transformative reform is narrow.
Policy Exchange
Policy Exchange has produced the most sustained conservative-reformist analysis of delivery failure. Its Reform of Government Commission, launched in 2020 and chaired by Dame Patricia Hodgson, produced Government Reimagined in May 2021, a comprehensive indictment of Whitehall's structural failures that reads, in retrospect, as a template for debates that followed.
Policy Exchange's distinctive contribution is the argument that capability matters more than scale. The state doesn't need more civil servants; it needs better ones. It doesn't need more consultants; it needs fewer, because their dominance is a symptom of in-house skill gaps. It doesn't need more regulation; it needs regulation that can be tracked, challenged, and removed. Its Rise of the Regulators (December 2024) documented 1,200 regulations introduced between 2014 and 2024, finding that fewer than 20% came with proper Impact Assessments and that the regulatory system was gripped by a "risk aversion ratchet."
The National Audit Office
The NAO is in a different institutional category from the others. It doesn't advocate. It audits. Its findings carry legal weight because they're based on access to government data that no think tank can replicate. And what the NAO's data shows about delivery is damning.
Three NAO reports are particularly significant for this synthesis. Digital Transformation in Government (March 2023) traced a consistent pattern of digital failure: decisions made before business problems are understood; capital funding allocated without revenue funding to operate new systems; legacy costs crowding out transformation budgets. Delivering Value from Government Investment in Major Projects (February 2024) examined the 244 projects in the Government Major Projects Portfolio and found that government rarely evaluates what happens after projects complete, whether they actually produced the value that justified their cost. Lessons Learned: Governance and Decision-Making on Mega-Projects (March 2025) analysed the £834 billion portfolio and found that mega-projects fail when design isn't mature before commitment, when no single person has authority, and when leadership churn prevents institutional knowledge from accumulating.
IPPR
IPPR approaches delivery failure from the progressive left. Its "Great Government" programme, launched in 2023, argues that New Public Management — the dominant reform paradigm since the 1980s — has run out of road. Importing private-sector targets and quasi-markets into public services has produced gaming, demotivation, and inequality. The new playbook needs to be built on high trust, high autonomy, and high skill.
IPPR's contribution is to frame delivery failure as a social contract problem, not just an administrative one. When government can't deliver, democratic legitimacy erodes. The Great Government series draws on international comparisons — Finland's education system, Dutch social care, UK devolution — to argue that better delivery is possible. The question is whether the institutional conditions can be created and sustained.
CPS and Reform
The CPS and Reform represent different strands of the right's case for delivery reform. CPS is primarily focused on fiscal sustainability and regulatory burden, approaching delivery through the lens of productivity and value for money. Its Future of Regulation (April 2024) documented £35 billion in gross annual regulatory costs imposed between 2010 and 2019 and described the entire regulatory reform apparatus as a "Potemkin system." Its Productivity Toolkit for Britain (February 2026) found public sector productivity still 5.7% below 2019 levels and identified specific levers — sick leave rates, estate utilisation, procurement waste — that could move the dial without additional headcount.
Reform is more focused on Whitehall culture and incentives. Its Efficiency Mindset (October 2023) argued that government lacks the performance management frameworks, incentives, and accountability structures to pursue efficiency gains. Its Making the Grade (April 2024) survey of 771 civil servants documented grade inflation, cognitive homogeneity, and specialist skill shortfalls. Its Getting the Machine Learning (September 2024) laid out why AI adoption was being squandered through the same institutional failures that have blocked previous technology waves.
Centre for British Progress
CBP occupies a distinctive niche. Where other organisations focus on Whitehall, CBP focuses on the physical infrastructure delivery system — planning, consenting, development corporations, transport investment. Its argument is that Britain's inability to build houses, roads, railways, and energy infrastructure at competitive speed and cost is itself a delivery failure, rooted in institutional design choices that systematically empower blocking over building.
CBP's publications on new towns, the Oxford-Cambridge corridor, AI Growth Zones, and transport devolution all make versions of the same argument: the UK's delivery system gives too many actors a veto and too few actors a mandate. Underground lines cost six times more per mile in Britain than in Spain. French trams cost half what British ones do. That's not a natural fact. It's a policy choice.
The Publication Trend
One thing the volume of output across these eight institutions confirms is that the delivery problem has been getting more, not less, urgent. Publication rates have accelerated. The IFG published more on delivery in 2024–25 than in any previous two-year period. TBI's output doubled between 2023 and 2025. CBP was founded specifically to address delivery gaps that other institutions were documenting but not solving.
This acceleration reflects both the worsening underlying data and a growing political appetite for answers. The 2024 general election brought a Labour government explicitly committed to "mission-driven" delivery. The think tank world mobilised around the question of whether missions could work in practice. The answer coming back, as of March 2026, is: not yet.
Part 2: The Six Areas of Consensus
The Centre of Government Is Too Weak to Drive Delivery
Strong consensusThe most important structural finding in the delivery literature is this: every prime minister since at least 1997 has wanted to deliver more than the institutional machinery of British government allows. The centre, as constituted, blocks delivery rather than driving it.
The IFG spent a year investigating this through its Commission on the Centre of Government, which reported in March 2024. The Commission's final report, Power with Purpose, was blunt: "The centre of government has failed to give successive prime ministers the strategic support they need to deliver their priorities." No.10, the Cabinet Office, and the Treasury are structurally misaligned — unclear roles, duplicated functions, insufficient strategic capacity, and fragmented accountability. The result is a government that is reactive rather than strategic.
The Commission's seven recommendations form the most comprehensive reform blueprint in the literature. They include: a new "Priorities for Government" framework agreed at the start of each parliament; an Executive Cabinet Committee of key ministers; a new senior first secretary of state for delivery; a restructured Department of the Prime Minister and Cabinet; a separate Department for the Civil Service; separated cabinet secretary and head of civil service roles; and a shared strategy, budget, and performance management process linking priorities to resources.
TBI reaches the same diagnosis from a different direction. Its foundational Governing in the Age of AI paper calls for an "AI Mission Control" at Number 10, headed by an "AI Mission CEO" with a mandate to drive cross-government transformation. The Chief Secretary to the Treasury should own AI transformation, with authority to direct departmental spending bids. The logic: without a powerful centre with real authority, AI transformation will fragment into departmental pilots that never connect.
TBI's Reimagining the Spending Review paper (June 2025) makes the institutional argument most explicitly: the existing Spending Review process is itself a barrier to delivery, driven by bilateral departmental bargaining that the centre cannot override. The missions framework, which Labour hoped would change this, "failed to meaningfully supplant entrenched patterns of departmental bargaining" in the 2025 Spending Review.
Policy Exchange's Government Reimagined reaches the same conclusion through its Reform of Government Commission work. The Prime Minister's office lacks the institutional machinery to force delivery across Whitehall. Extended Ministerial Offices should be restored to give secretaries of state access to expert advisers. Permanent Secretary tenures should be made conditional on delivery track records.
The IFG's Mission Launch paper (September 2024) found that two months into the Starmer government, Labour had made commitments but not established the institutional architecture needed to deliver them. It proposed a Mission Strategy Board chaired by the PM, lead cabinet ministers for each mission, and mission strategy teams with cross-departmental authority. The subsequent Whitehall Monitor 2026 found these recommendations largely unimplemented: mission governance had "collapsed" into a confusing array of rebranding exercises, and 2025 had been "missed" as a reform window.
IPPR's Mission-Driven Government paper (May 2024) contributed interview evidence from over 30 experts to the same finding: missions require institutional equivalents of the 1997 reforms — spending reviews, delivery units, outcome frameworks — but calibrated to the complexity of cross-cutting missions that can't be reduced to simple metrics. The feedback loop between policy design and frontline delivery needs to be rebuilt, not patched.
The strength of this consensus is remarkable. The IFG and Policy Exchange don't agree on much. TBI and Reform approach government reform from very different assumptions. IPPR and CPS disagree about the size of the state. But they all agree that the current centre of government can't drive delivery, and they all propose versions of the same fix: give the centre more authority, more strategic capacity, and more real accountability for outcomes.
What the consensus doesn't contain is a convincing account of how these reforms survive the change of government. The Blair PMDU, the gold standard of UK delivery architecture, was abolished by Cameron in 2010. Oliver Letwin later called this "a terrible mistake." The Thatcher efficiency programme was diluted by Major. Every reform of the centre has been restructured or reversed within two parliaments. The consensus exists; the institutional protection doesn't.
Civil Service Capability Has Eroded
Strong consensusThe civil service has grown. It has not improved. That tension runs through the entire delivery literature, and the data is cross-partisan and consistent.
The IFG's Whitehall Monitor series tracks the numbers annually. The civil service grew from under 380,000 to 520,440 FTE between 2016 and 2025. Grades 6 and 7 rose 121% since 2010. Turnover reached 12.7%, the second highest since 2010. Median civil service pay is still slightly below its 2010 real-terms level despite 2024 pay rises. Consultancy and temporary staff spending rose 40% in real terms since 2018/19 — a direct measure of the in-house capability gap.
The capability picture is documented in detail by the IFG's Who Runs Whitehall report (May 2024). Of 148 permanent secretaries and directors general studied, 52% attended Oxbridge. Only 23% have STEM degrees. Only 12% are based outside London. Policy professionals are "numerically and culturally dominant," making up 43% of the senior tier. Only 22% of permanent secretaries have had leadership roles outside government for more than three years.
The Opening Up report (December 2022) found that only 18% of new entrants to the senior civil service in 2020/21 were external recruits. The IFG's 20 Ways to Improve the Civil Service (August 2024) synthesised over a decade of research into concrete, legislation-free actions: enforce minimum terms of service, advertise all jobs externally by default, set up large-scale secondment programmes, create senior specialist roles.
Policy Exchange's Government Reimagined identified the same capability gaps in 2021 and proposed specific fixes: a new pay grade above the SCS to attract specialists; dramatically more Senior Responsible Owners for major programmes; mandatory training on procurement, project management, and digital delivery for ministers and permanent secretaries.
Its Smaller, Better, Higher Paid? paper (May 2025) updated the diagnosis. Since 2016, the civil service expanded from under 380,000 to over 514,000 by 2024. The grade inflation, rapid churn, uncompetitive specialist pay, and excessive management layers serve nobody — not service users, ministers, or civil servants themselves. The paper proposes £5 billion in savings through structural delayering, recycling money into competitive pay for digital, commercial, and scientific roles.
Reform's Making the Grade survey of 771 civil servants found that grade inflation has produced excessive management layers, the fast stream has become cognitively homogeneous in a way that generates groupthink, and specialist skills remain chronically undersupplied. Its Manifesto for Delivery (June 2024) found civil servants themselves identifying the same problems: cultural premium on policy over delivery expertise, inadequate cross-departmental collaboration mechanisms, weak evaluation and learning systems.
TBI's analysis of skills from a technology angle reaches the same conclusion. AI transformation requires civil servants who can write problem definitions, manage vendors, act as intelligent clients for technical work, and interpret data outputs. These skills are systematically absent. Governing in the Age of AI proposes benchmarking AI-related civil service salaries to at least 75% of private-sector market rates. Delivering AI Impact (February 2026) identifies the binding constraint on AI transformation as "institutional and political, not technological" — the skills and incentive problem, not the tool problem.
The IPPR's Mission-Driven Government argues that an activist industrial state requires capabilities that the current civil service doesn't possess — co-design, partnership, and analytical skills suited to running complex multi-sector programmes, not just commissioning and regulating.
The NAO's Improving Operational Delivery in Government (March 2021) identified operational management capability — not policy design — as the critical limiting factor on service quality. Senior leaders lack the frameworks, incentives, and skills to manage service delivery effectively. That finding, made in 2021, has been confirmed repeatedly by the data since.
There is one area of genuine divergence within this consensus. CPS and Policy Exchange argue that capability problems are partly caused by a civil service that is too large — that grade inflation and management bloat crowd out specialisation. IFG and IPPR are more agnostic on size, focusing instead on the composition and skills of whoever is there. Reform sits closer to the CPS/Policy Exchange view on workforce structure, while agreeing with IFG on the cultural dimension. This doesn't crack the consensus — they all agree capability has eroded — but it matters for the remedy.
Major Project Delivery Is Structurally Broken
Strong consensusThe data on major project delivery is the most alarming in the entire delivery literature, and it's the area where the cross-institutional consensus is most complete. Britain can't build things. The evidence is not contested.
The Government Major Projects Portfolio contained 227 projects as of March 2024, with a combined whole-life cost of £834 billion, according to the NAO's Lessons Learned: Governance and Decision-Making on Mega-Projects (March 2025). Of those, only around 11% carry a "green" Delivery Confidence Assessment from the Infrastructure and Projects Authority, meaning on track to succeed. TBI's Governing in the Age of AI cites the same figure: only around 10% of government major projects have a green rating, while 183 are rated amber.
The IFG's Major Projects in Government explainer found that in 2020, 11 projects in the portfolio (9%) were rated "unfeasible" — a record high, up 6 percentage points from 2019. Only 3 of 123 projects with known ratings (2%) were rated "highly likely" to succeed on time and on budget. HS2 was rated "unfeasible" for the first time in 2020, with whole-life costs of £55.7 billion. By the time Phase 2 was cancelled, it had become a case study in premature commitment and inadequate design maturity.
The NAO's Delivering Value from Government Investment in Major Projects (February 2024) went further than the confidence ratings. It found that government rarely evaluates what happens after major projects are completed. It tracks delivery to budget and schedule, but not whether projects produced the value — for citizens, service users, or the environment — that justified their cost. By examining six completed projects including High Speed 1 and the Diamond Light Source, it identified seven factors that determine whether major investments create or destroy value. A clear vision of intended benefits, maintained through design decisions, is the most important. It's the one most commonly absent.
The NAO's megaprojects report identified the structural causes of failure with precision: designs and consents committed to before they're mature; no single authoritative decision-maker; leadership churn that prevents institutional knowledge from accumulating. The report quotes the Institution of Civil Engineers directly: be "ruthless at gate reviews and controls until design and planning are sufficiently mature." Immature designs are the primary driver of cost escalation. The UK consistently doesn't wait.
CBP's analysis of infrastructure delivery costs makes the international comparison explicit. Let Mayors Build (October 2025) found that underground lines in Spain cost one-sixth as much per mile as in Britain and French trams cost half as much as British ones. This isn't a natural feature of geography or geology. It's the product of incentive structures: when projects are funded with national money, localities have no reason to control costs; when approval is controlled centrally, there's no pressure to move quickly. The institutional design produces the cost premium.
CBP's Project Hawking report (December 2025) on the Oxford-Cambridge corridor identifies the same failure mode at scale. Incremental interventions — Development Consent Orders, local planning decisions, piecemeal development corporations — can't deliver at the speed and scale the corridor needs. The "unholy trinity" of local politics, dysfunctional regulation, and insufficient financing blocks every approach. The solution CBP proposes is radical: a single development corporation with supreme planning authority, outside normal government processes, self-funding through land value capture.
Policy Exchange's Government Reimagined identified the shortage of capable Senior Responsible Owners as a critical delivery bottleneck, arguing that £600 billion in gross public sector infrastructure investment cannot be managed with the current SRO cadre. The NAO's megaprojects report and Reform's Manifesto for Delivery reach the same conclusion independently: senior responsible owners need to stay in post longer, and their accountability needs to be tied to delivery outcomes, not just business case approval.
The NAO's Government Shared Services reports (November 2022 and March 2026) document the same failure pattern applied to internal reform programmes. Since 2004, central government has repeatedly tried to cut costs by sharing back-office functions across departments. Every strategy has failed to deliver intended savings. The refreshed 2021 strategy, grouping departments into five cloud-based shared service clusters by 2028, is similarly at risk. The fundamental enablers of process standardisation and data convergence have not been put in place. Four years in, the pattern of the previous 17 years is repeating.
The defence procurement picture adds another dimension. CBP's Fixing UK Defence Procurement (December 2025) identifies four systemic failures in defence major programmes: scope creep and over-specification; inter-service rivalry creating too many decision-making layers; feast-and-famine ordering patterns that hollow out sovereign industrial capacity; and procurement cycles too slow to adopt software-driven, AI-enabled capabilities. These failures are structural, not accidental. They're built into the institutional architecture.
Digital and AI Transformation Is Necessary but Under-Resourced
Moderate consensusThere's a number that now anchors the digital delivery debate: £40 billion. TBI's Governing in the Age of AI estimates that AI could generate £40 billion per year in public sector productivity gains, totalling £200 billion over five years. IPPR's Great Government estimates £24 billion per year. CPS and Reform reference the government's own internal estimate that a third of civil service tasks could be automated. None of these figures are identical, but they're all large enough to change the fiscal position fundamentally. The institutions agree on the scale of the prize.
They also agree that the prize isn't being captured. The NAO has tracked this most systematically. Its Digital Transformation in Government report (March 2023) found that despite the creation of the Central Digital and Data Office in 2021 and a new Digital and Data Roadmap, government's track record of digital transformation remained poor. Technology decisions are taken too early, before business problems are properly understood. Capital funding is secured without the revenue funding to operate new systems. Legacy systems consume an ever-growing share of digital budgets, crowding out transformation.
The NAO's Government's Approach to Technology Suppliers (January 2025) found that government spends at least £14 billion annually on digital procurement, with £3 billion in cost increases identified across just five digital programmes. Commercial teams in departments lead procurement decisions without adequate digital expertise. Post-contract supplier management is almost entirely absent. The centre has not aligned responsibilities, skills, or resources to lead digital procurement strategically.
TBI's departmental case studies bring this to life. Its DWP analysis found that DWP could free up 40% of its time using AI tools, equivalent to nearly £1 billion per year. DWP fraud and error cost close to £9 billion per year. The tools to address this exist. The deployment architecture doesn't. Its NHS analysis found that without a universal digital health record as foundational infrastructure, the NHS can't unlock AI-era benefits because health data remain fragmented across GP practices, hospitals, and pharmacies.
TBI's Reimagining Local Government (May 2025) quantifies the local dimension: AI could automate or improve 26% of local government tasks, equivalent to 1 million hours per year in a single council, or £8 billion nationally. A 227,000-person care assessment backlog could be cleared in one month with AI-enabled processing. Only 20% of major planning applications are settled within the statutory timeframe. Seven in ten councils cite high-quality data availability as a barrier to innovation.
IFG's assessment of digital government is notably more cautious. Its Digital Transformation and AI chapter in Whitehall Monitor 2025 noted that the civil service is taking AI more seriously than before, with encouraging pilots and the consolidation of digital teams in DSIT, but that legacy IT systems remain a serious problem and the government lacks a clear roadmap that addresses underlying fragmentation. Its January 2026 commentary on the government's digital roadmap said it "points to an appealing future but fails to set out how to get there."
Reform's Getting the Machine Learning identified why AI adoption specifically is being squandered: there is no single point of leadership, no systematic approach to high-value use cases, no cross-government procurement framework, and no mechanism for scaling successful pilots. The creation of the digital centre of government in DSIT provides an opportunity to make these changes, but only if DSIT is given genuine authority, not just a title.
The 2025 Spending Review allocated significant sums — £10 billion to digitalise the NHS, £1.2 billion for cross-government digital, £2 billion for the AI Opportunities Action Plan. But TBI's Unpacking the 2025 UK Spending Review noted that the Health Foundation estimates NHS and adult social care digitalisation actually requires £21 billion over five years, significantly more than allocated. The gap between what's needed and what's funded is the story of UK digital government for three decades.
Accountability Mechanisms Have Become Obstacles to Delivery
Strong consensusThis is the most politically charged of the six consensuses because it cuts directly against the instincts of both the parliamentary left (which values public inquiries as accountability tools) and the legal and professional establishment (which values judicial review and impact assessment requirements). But the evidence base for it is now substantial and cross-partisan.
The IFG's How Can the Government Ensure Inquiries Are Set Up to Succeed? (December 2025) found that 27 public inquiries were underway simultaneously in December 2025, a record. Seven new ones were announced in 2025 alone. The Covid-19 Inquiry alone has cost the government £100 million in response costs, including £56.4 million in legal costs. Inquiries are taking longer, costing more, and too often failing to deliver meaningful change. The underlying drivers include low public trust pushing demand toward statutory inquiries, an absence of systematic knowledge-sharing, and risk aversion discouraging innovation in how inquiries are run.
Policy Exchange's Rise of the Regulators documents the regulation side of the same problem. Based on analysis of over 1,200 regulations introduced between 2014 and 2024, it finds that the UK's regulatory state is gripped by a "risk aversion ratchet" that generates new rules and almost never removes old ones. The result is a productivity-suppressing ceiling that is diverting doctors, nurses, and police officers from core duties. Natural England's nutrient neutrality requirements alone paused 150,000 homes. The regulatory expansion represents a democratic deficit: elected lawmakers outsourcing executive power to agencies with little parliamentary accountability.
The CPS's Future of Regulation takes this further. The entire regulatory reform apparatus is described as a "Potemkin system" — Impact Assessments produced by junior civil servants late in the day to justify decisions already made, read by almost no ministers, subject to systematic measurement errors. The system creates the appearance of accountability without delivering it.
CBP's infrastructure analysis shows what this means in practice. Its Unblocking AI Growth Zones paper (October 2025) found that the planning and environmental regulatory requirements that blocked energy, housing, and transport projects will equally block AI datacentre development if not addressed. The typical five-year approval timeline for nationally significant infrastructure projects is a delivery killer. Its New Towns for a New Generation work found that just one in five major planning applications are decided within the 13-week statutory period and that costs of obtaining permission had risen more than 30% for half of SME builders since 2021.
Reform's productivity analysis confirms the cost. The State of the State 2025 survey of public sector leaders found that more directive central leadership is needed to accelerate digital transformation — the current devolved, departmental approach is too slow. Leaders express concern about governance mechanisms. The compliance burden — the Social Value Act, the Public Sector Equality Duty as applied to procurement, box-ticking requirements — diverts resources from actual delivery.
The IFG's Improving Accountability in Government Procurement (September 2024) found that the Procurement Act 2023 improved transparency requirements, but that implementation risks falling short because departments lack commercial capability to set meaningful KPIs. The accountability architecture creates the appearance of oversight without the substance.
What these institutions agree on is that the UK's accountability architecture produces more process than accountability. Inquiries that take years and cost millions but rarely change policy. Impact Assessments that are written after decisions are made. Judicial review processes that allow strategic delay of nationally significant decisions. Procurement rules that prioritise compliance over outcomes. Environmental regulations that apply serially rather than being resolved strategically. The consensus is that this system doesn't work, but the institutions disagree sharply on what to do about it — which is why this consensus is politically charged without generating a coherent reform programme.
The Treasury Controls Too Much and Delivers Too Little
Strong consensusThe final area of consensus is about fiscal architecture. The Treasury, in its current form, is a barrier to delivery. That's not a claim the Treasury makes about itself. But it's a claim that emerges from across the spectrum.
The IFG's How the Centre of Government Can Design Better Growth Policy (January 2026) argues that the UK's chronic growth failure is substantially a governance failure — the centre of government lacks the strategic capacity, institutional authority, and cross-departmental co-ordination needed to design and sustain coherent growth policy. The Treasury is part of this problem. Its departmental bilateral spending deals reinforce silos and prevent cross-cutting missions from generating their own funding streams.
TBI's Reimagining the Spending Review (June 2025) is the most detailed critique. The UK Spending Review process is itself a delivery barrier — bureaucratic, opaque, driven by bilateral departmental bargaining, and fundamentally misaligned with outcomes-focused cross-cutting missions. It proposes replacing the SR with an AI-powered "Strategic Review" — a real-time portfolio-management system that funds what works and reallocates from what doesn't. Its analysis of the SR 2025 found that the missions approach "failed to meaningfully supplant entrenched patterns of departmental bargaining" — the Treasury's institutional culture absorbed the missions framework rather than being changed by it.
IPPR's Great Government: Public Service Reform in the 2020s argues for long-term cross-sector missions with independent accountability bodies modelled on the Climate Change Committee, precisely because this takes accountability out of the departmental bilateral structure and places it in an independent institution that can't be negotiated away in a spending round.
IFG's Austerity Postponed? assessment of Labour's first budget noted that while day-to-day spending rose substantially in 2024/25 and 2025/26, spending growth slows sharply after that, implying further cuts to unprotected areas. The Treasury's medium-term fiscal position continues to squeeze the very services and reforms the government says are its priorities. The fiscal architecture and the delivery architecture are misaligned.
The IFG's Whitehall Monitor 2026 found that the 2025 Spending Review "reinforced departmental silos rather than enabling cross-mission working." It failed to create shared mission budgets despite the rhetoric. The government needs to shed 29,000–40,000 civil servants by 2029/30 to meet admin budget targets, but has no credible workforce plan for doing this.
The CPS and Reform approach the Treasury's role from a different angle, focusing on how Treasury rules make efficiency savings difficult to sustain. Reform's Efficiency Mindset argues that departments should keep a share of any efficiency savings they generate, rather than having these automatically recaptured by HMT. The current system creates a perverse incentive: why pursue efficiency if the Treasury pockets the gains?
Policy Exchange's Government Reimagined recommends reforming the Green Book and procurement rules to make long-term investment decisions less distorted by Treasury discount rate assumptions and short-term value-for-money tests. Infrastructure with long payback horizons is systematically undervalued by the current analytical framework.
The consensus is that the Treasury, as currently designed, optimises for fiscal control rather than for delivery outcomes. It's good at stopping money being spent; it's not good at ensuring that money spent achieves what it's supposed to. Cross-cutting missions require cross-cutting budgets. The Treasury's bilateral structure prevents both.
Part 3: The Three Areas of Genuine Dispute
Should the State Get Bigger or Smaller?
ContestedThis is the sharpest ideological fault line in the delivery literature, and it's genuinely a philosophical divide, not a manufactured one.
IPPR and TBI start from the premise that the UK state is under-resourced for the tasks it's being asked to perform. IPPR's Great Government argues that public services require more investment, but investment structured around reform — prevention over acute services, devolution over central command, learning ecosystems over target-setting. TBI goes further: its Economic Case for Reimagining the State (July 2024) notes that government spends almost 10% of national income on public sector wages and needs to invest £4 billion per year in AI rollout costs this parliament to unlock the productivity gains that would eventually reduce that share. The immediate prescription is spend more to spend less later.
IFG's position is closer to neutral on size, but its data points consistently to under-resourcing. The Performance Tracker series documents services that have been starved of capital investment for decades — the £37 billion maintenance backlog across schools, hospitals, prisons, courts, and roads documented in its 2024 pre-election report. This isn't a case for a larger state per se, but the data shows that the existing state can't function adequately on existing resources.
CPS, Policy Exchange, and Reform argue the opposite. Policy Exchange's Smaller, Better, Higher Paid? makes the case explicitly: the civil service has expanded by 135,000 FTE since 2016 without a corresponding improvement in delivery. The expansion has produced more management, more grade inflation, more process, and less capacity to actually do things. A smaller, better-paid, more specialist civil service would deliver more. The paper proposes £5 billion in annual savings through structural delayering.
CPS's Productivity Toolkit focuses on the NHS specifically: government spending plans for health are predicated on NHS productivity improving at four times its recent historical average. If this doesn't happen, the Chancellor will need an additional £20 billion for the NHS by 2028-29. The implication is that more spending without structural reform doesn't solve the delivery problem — it just moves it further down the road.
Reform sits between these positions in an interesting way. Its Efficiency Mindset argues that the question of size is less important than the question of culture: a large civil service with strong performance management and efficiency incentives would be better than a small one without them. Its State of the State 2025 (with Deloitte) found public sector leaders calling for more directive central leadership to accelerate change, which implies a more effective centre, not necessarily a smaller one.
The dispute matters for the delivery literature because it shapes what reforms are even attempted. If you believe the state needs more capacity, you propose more institutions, more civil servants with better skills, more investment. If you believe the state needs to do less, you propose cuts to management layers, regulatory rollback, and concentration of effort on genuinely critical priorities. Both positions have evidence behind them. Neither has been tested at sufficient scale in the UK to produce a decisive verdict.
What neither side has resolved is the sequencing problem. Even those who believe the state ultimately needs to be smaller accept that building specialist capability takes time and investment. Even those who believe the state needs more resource accept that additional spending on the existing system without reform is wasteful. The practical gap between "invest to reform" and "reform to invest" is narrower than the rhetoric suggests, but the political and fiscal framing is very different — and that framing shapes what's politically achievable.
Centralise or Devolve?
ContestedThe second genuine dispute is about architecture: should delivery be driven from the centre, or should it be handed to local and regional institutions with the mandate and resources to do it themselves?
TBI's reform agenda is explicitly centrist. Its Governing in the Age of AI calls for a powerful AI Mission Control at Number 10, a Chief Secretary to the Treasury directing departmental technology spending, and a "Bezos mandate" requiring all departments to open their APIs. Its Time for Digital ID (September 2025) proposes a dedicated Digital ID Delivery Unit backed by the Prime Minister's direct authority. The logic is consistent: transformation at scale requires central authority. Without it, departments fragment into incompatible silos and pilots never become systems.
IFG broadly agrees that the centre needs strengthening, but its Commission on the Centre of Government simultaneously recommends devolution as a necessary complement. Its Case for Total Place 2.0 (May 2025) argues for 10–15 pilot areas with genuine integration of health, social care, housing, and criminal justice spending — a fundamentally local approach to cross-cutting delivery. Its Do the Government's Public Service Reform Plans Add Up? (March 2026) warns against "re-centralisation" undermining the government's stated commitment to devolution and integration.
IPPR is the most clearly pro-devolution voice in the literature. Its Mission-Driven Government explicitly calls for more powers for combined authorities. The Great Government playbook draws on Finnish education and Dutch social care — both highly devolved, high-trust models — as the template. The argument is that frontline autonomy and local flexibility are the conditions for the intrinsic motivation that makes reform sustainable.
CBP represents a third position that doesn't fit neatly into either camp. Its development corporation model — the Hawking DevCo, the AI Growth Zone Authority, the New Towns development corporations — bypasses both central government processes and local government structures. These are purpose-built delivery institutions with statutory authority, operating at the scale of a region or corridor, accountable to Parliament for outcomes rather than to any existing tier of government. They're neither central nor local in the conventional sense.
CBP's Let Mayors Build advocates devolution of transport funding to metro mayors, but on formula-based terms that give mayors genuine fiscal incentives to control costs — not the current system of competitive bidding for project-specific national grants that gives mayors no reason to care about value for money.
Policy Exchange is genuinely divided internally on devolution. Its civil service reform work favours centralising some standards and recruitment processes while its planning and housing work recognises that local political opposition is a primary delivery bottleneck. Its Broken Housing Market paper calls for a rules-based planning system with a presumption in favour of development, which is effectively a form of centralising the policy while decentralising execution.
The dispute is genuine because both centralisation and devolution address real problems. Centralisation enables standards, prevents lowest-common-denominator compromises, and allows scarce specialist capacity to be shared. Devolution enables local ownership, prevents one-size-fits-all failure modes, and allows services to be shaped to local needs. The UK's current system combines the costs of both: centralised enough to prevent local innovation, devolved enough to prevent national standards. The debate about which direction to move is real, and the institutions diverge meaningfully on the answer.
Technology as Saviour or Supplement?
ContestedThe most forward-looking dispute in the literature is about the role of technology. TBI is the most techno-optimist institution in the field. IFG is the most cautious. NAO's audit record is the most sobering. Reform sits between them. The gap between these positions matters because it shapes how urgently transformation is pursued and how much institutional change is required to support it.
TBI's position is the boldest. Its Public-Service Reform in the Age of AI (January 2026) proposes a complete reimagining of how government operates: three new infrastructure layers (universal digital ID, modular public-service platforms, a real-time system-intelligence layer), a shift from command-and-control to "signal and steer" governance, and the replacement of locked institutional budgets with modular, outcome-based funding. This isn't incremental digitisation. It's a claim that the current model of government — bureaucratic, labour-intensive, standardised, reactive — will be replaced by something qualitatively different. The DWP case study shows this in concrete terms: AI tools could free up 40% of DWP staff time, clear the PIP backlog in 10 months, and reduce the £9 billion per year fraud and error bill.
IFG's assessment is more measured. Its January 2026 commentary on the government's digital roadmap found it "points to an appealing future but fails to set out how to get there." Its Whitehall Monitor 2025 chapter on digital and AI noted encouraging pilots and welcome consolidation of digital teams in DSIT, but warned that legacy IT systems remain a serious problem and that the government lacks a clear implementation plan. The subtext of much IFG digital analysis is consistent: the technology isn't the hard part. The institutions, the incentives, the skills, and the procurement architecture are the hard part. Fixing those first would allow almost any technology to work. Not fixing them means the best technology will produce the same results as everything before it.
NAO's audit record is the most sobering. Its Digital Transformation in Government (March 2023) traced a consistent failure pattern across two decades of digital government initiatives: wrong sequencing, capital without revenue funding, legacy costs dominating budgets, supplier dependency, and absent post-contract management. The Government's Approach to Technology Suppliers (January 2025) found £3 billion in cost increases across just five digital programmes, and a government that still doesn't know accurately how much it spends on digital procurement. This isn't a case against digital transformation — NAO explicitly calls for it. It's evidence that the current institutional architecture can't deliver it.
Reform's position in Getting the Machine Learning acknowledges the transformative potential of AI while focusing on why adoption isn't happening: no single point of leadership, no framework for high-value use cases, no cross-government procurement structure, no mechanism for scaling pilots. Reform is broadly pro-technology, but its analysis suggests that the technology optimism of TBI needs to be grounded in the institutional realism of IFG and NAO before it can translate into delivery.
TBI itself recognises this tension in some of its work. Its Delivering AI Impact (February 2026), focused on international examples, explicitly says the binding constraint on AI transformation is "institutional and political, not technological." Its Reimagining Local Government paper notes that seven in ten councils cite high-quality data availability as a barrier to innovation — a data governance problem, not a technology problem. There's a version of TBI's argument that is compatible with IFG's caution: the technology is necessary but the institutions have to change first.
The dispute matters because it shapes the policy sequencing. If technology is the primary lever, you prioritise digital investment and AI procurement, and you accept that institutional reform will follow. If institutions are the primary lever, you fix the centre of government, the SRO cadre, the procurement architecture, and the skills pipeline first, then invest in technology on top. The two approaches aren't mutually exclusive, but they're different in emphasis, and the difference is real.
Part 4: What No One Is Saying
The 101 publications reviewed here represent the most comprehensive body of analysis on UK government delivery ever produced. They contain important insights, strong evidence, and genuine proposals. They also contain some significant gaps.
No institution has a convincing theory of political protection for reform.
The Blair PMDU was abolished by Cameron. Oliver Letwin called this "a terrible mistake." The Coalition's efficiency reforms were diluted by Cameron and then effectively abandoned by May. The Levelling Up agenda, such as it was, collapsed within two years of the legislation that created it. HS2 Phase 2 was cancelled after decades of commitment. The pattern is not random: UK government reform is structurally vulnerable to political change. When leadership changes, reforms that were "owned" by the previous administration are restructured or reversed — sometimes for good reasons and sometimes for bad ones.
The institutions have documented this pattern. None of them has proposed a convincing solution to it. The IFG's Commission on the Centre recommends new structural arrangements for No.10 and the Cabinet Office, but these could be reversed by any incoming prime minister. TBI's delivery unit proposals are vulnerable to the same charge. IPPR's independent accountability bodies modelled on the Climate Change Committee are the closest the literature gets to institutional protection, but IPPR applies this specifically to public service missions, not to the delivery architecture itself.
The Climate Change Act model is instructive precisely because it's the exception that proves the rule. The Act locked in carbon targets in statute, created an independent committee to monitor progress, and required the government to explain publicly why it was deviating from its own trajectory. Every subsequent government has maintained it, not because it was politically convenient to do so, but because the costs of departing from it — reputational, legal, and political — were too high. The question of whether something analogous could protect delivery architecture has barely been asked.
Almost no one is working on delivery culture.
The literature is full of proposals for delivery structures: mission boards, delivery units, development corporations, SRO tenures, procurement frameworks. It's almost silent on delivery culture — the attitudes, incentives, and working norms that determine how the people inside those structures actually behave.
Reform's Efficiency Mindset is a partial exception, arguing that Whitehall lacks the efficiency culture that would make structural reforms effective. IFG's Policy Making in the Real World (December 2024), based on 50 interviews with senior civil servants and 20 former ministers, found that reforms to process, qualities, structures, and politics all fell short because they failed to account for the role of political incentives. But this is rare in the literature. Most institutions focus on org charts and financial flows, not on the lived experience of people trying to deliver things inside a system that doesn't reward delivery.
The gap matters because structural reforms don't change culture automatically. The PMDU had the right structure and created substantial culture change in departments that it worked with closely — and left much of Whitehall untouched. Mission boards, however well designed, will be staffed by civil servants whose career progression depends on avoiding failure rather than achieving success. Until the promotion criteria, the appraisal systems, and the informal norms around risk-taking are addressed, structural reform will have limited impact on the actual experience of delivering.
The incentive structure for risk aversion has been under-analysed.
Why are civil servants risk-averse? The question seems obvious, but the answer is more specific and addressable than the general culture diagnosis suggests. Accountability for failure is immediate and public — select committees, NAO reports, ministerial blame. Accountability for success is delayed and diffuse — outcomes emerge years after decisions. The career costs of a visible failure are high; the career rewards of a quiet success are low. This isn't a character trait of civil servants. It's a rational response to the incentive structure they operate in.
Almost none of the institutions have proposed directly rewiring these incentives. The IFG's 20 Ways to Improve the Civil Service addresses tenure and pay, but not the fundamental accountability asymmetry. Reform's efficiency paper proposes that departments keep a share of efficiency savings — a partial incentive fix. But the core problem — that the system creates stronger disincentives for failure than incentives for success — hasn't been addressed by any institution with a comprehensive proposal.
No institution has seriously proposed statutory delivery targets applied to public services.
IPPR's Great Government: Public Service Reform in the 2020s proposes legislating new national missions with independent accountability bodies modelled on the Climate Change Committee. This is the closest the literature gets. But it applies the model to outcome missions (the healthiest country, the safest country), not to delivery mechanisms themselves. No institution has proposed a statutory framework that requires the government to report independently verified delivery progress against a fixed schedule, with legal consequences for persistent failure.
The Climate Change Act model works because it bites. The consequences of missing carbon budgets are political and legal, not just reputational. There's no equivalent in public services. Waiting list targets, crime reduction commitments, housing delivery promises — all can be quietly dropped, redefined, or blamed on external factors without triggering any formal accountability mechanism. The institutional architecture for holding governments to their own delivery commitments simply doesn't exist.
The local government delivery gap is under-researched.
The IFG has done useful work on local government funding through its Neighbourhood Services Under Strain analysis and its local government sections in the Performance Tracker. CBP has focused on infrastructure delivery at local level. TBI's Reimagining Local Government provides a technology lens. But what's missing is a comprehensive analysis of local government as a delivery institution: its capability, its accountability architecture, its relationship with central government, and its capacity to absorb the devolution that multiple institutions are calling for.
Seven councils have effectively gone bankrupt since 2018. Section 114 notices (council effective bankruptcies) appeared six times more frequently in six years than in the previous three decades. Adult social care has a 227,000-person backlog. Children's services are overwhelmed. And yet the delivery literature on how local government actually operates — what determines whether a council is effective at delivery, and how capability can be built and maintained — is thin. If devolution is part of the answer to the central delivery problem, the question of local government delivery capacity deserves more rigorous treatment than it's currently receiving.
Part 5: Blueprint's Assessment
The Diagnosis Is Settled
The first and most important thing to say is that the argument has been won. You don't have to take Blueprint's word for it that the British state can't reliably implement policy. You can read the IFG's Performance Tracker, which shows eight of nine public services performing worse in 2024 than in 2010. You can read the NAO's major projects data, which shows only 11% of the Government Major Projects Portfolio rated green. You can read TBI's productivity analysis, which shows inputs to public services rising 25% between 2019 and 2024 while output grew 14%. You can read Reform's civil service surveys, Policy Exchange's regulatory burden analysis, and CPS's productivity data.
These are not marginal institutions with fringe views. They span the full width of the British political spectrum. They compete for influence, disagree on policy, and serve different parts of the political coalition. What they don't disagree on is that the state cannot reliably deliver what it promises — and that this failure is structural rather than incidental.
This matters because a political debate that used to be about whether the problem exists can now be about what to do. The institutions have collectively moved the Overton window on state capacity from a specialist concern to an organising question for British politics. That's a significant shift, and Blueprint has been part of producing it.
The Remedy Requires Institutional Protection
The second conclusion is harder. Even if we know what works — tightly focused delivery units with prime ministerial authority, capable SROs with long tenure, cross-departmental mission budgets, specialist skills built in-house, digital infrastructure properly sequenced — we don't have a reliable way of making these things stick.
The history documented in this synthesis is a history of reforms adopted and then reversed. Blair's PMDU was abolished in 2010. The coalition's efficiency programme was diluted. Mission boards were introduced in 2024 and had "collapsed" into rebrand territory by 2026, according to IFG's own assessment. HS2 Phase 2 was cancelled after decades of commitment. The pattern is not bad luck. It's a structural feature of Westminster government: each administration inherits the apparatus of the previous one and has both the motivation and the constitutional authority to reshape it to its own priorities.
This is why Blueprint's core thesis — the Gap Tracker — focuses on institutional permanence. Reforms that can be reversed by the next government aren't really reforms. They're demonstrations of intent, some of which happen to produce results before they're undone. What the delivery system needs is architecture that survives changes of government: statutory, independently monitored, with real consequences for failure.
The Case for Statutory Delivery Architecture
The Climate Change Act 2008 is the model. It passed with near-unanimous support. It created an independent Climate Change Committee to set carbon budgets and monitor progress. It required the government to explain publicly when it was deviating from its own trajectory. It survived changes of government. It has been strengthened, not weakened, over time. And it created the conditions for sustained, cross-party progress on an enormously difficult delivery challenge.
There is no equivalent in public services. Waiting lists can be redefined. Housing targets can be quietly dropped. Crime reduction commitments disappear. When the NAO reports failure, the government explains, and the cycle continues. The accountability architecture creates the appearance of scrutiny without the substance of consequences.
Blueprint's proposal for a permanent delivery tracker takes the Climate Change Act model seriously. It would require the government to publish, at fixed intervals, independently verified data on delivery progress across a defined set of public service commitments. It would create an independent body with statutory authority to assess that data and publish findings. It would require the government to explain publicly when it deviates from its own trajectory. It would not remove ministerial discretion — governments would still be able to change course. But they would have to do so transparently, against a baseline that can't be quietly manipulated.
IPPR's Great Government: Public Service Reform in the 2020s proposed independent accountability bodies modelled on the Climate Change Committee for each mission. This is directionally right but insufficiently focused on the delivery mechanism. The accountability body shouldn't just track outcomes. It should track the delivery commitments that are supposed to produce those outcomes — the SRO appointments, the programme timelines, the milestone schedules, the budget allocations. That's what a delivery tracker does.
The Technology Argument Is Necessary but Not Sufficient
TBI's case for AI-driven public service transformation is compelling. The potential gains are real. The DWP case study is persuasive. The NHS digital health record proposal is well-argued. The £40 billion per year estimate, even if it's an upper bound, represents a scale of opportunity that justifies significant investment and institutional energy.
But IFG and NAO are also right. The technology isn't the hard part. The institutions are. A government that can't maintain SRO tenure for major projects, that can't sequence digital investments correctly, that can't manage strategic suppliers as an intelligent client, and that doesn't have revenue funding to operate new systems it builds — that government will not capture the AI productivity prize no matter how good the technology gets.
The right sequencing is: fix the institutional architecture first, then use it to deploy technology at scale. That means a centre of government that can drive cross-departmental change, a civil service with the specialist capability to act as intelligent clients for technology procurement, a procurement framework that can run iterative test-and-learn approaches rather than waterfall contracts, and a legacy systems replacement programme with genuine political backing and dedicated resource.
TBI's Delivering AI Impact says this directly, even if it's not the headline message: "the binding constraint is institutional and political, not technological." Blueprint agrees. The technology argument is a reason to get the institutions right urgently, not a reason to hope that technology fixes broken institutions.
The Consensus Is Strong Enough to Act On
The final conclusion is the most important one. The obstacle to better delivery isn't ideas. It's not a lack of evidence. It's not a missing proposal or an unidentified mechanism.
Every institution covered in this synthesis — from CPS on the right to IPPR on the left, from TBI inside the current government's intellectual orbit to NAO as independent auditor — agrees on the core diagnosis. The centre is too weak. Civil service capability has eroded. Major projects are structurally broken. Digital transformation is necessary and under-resourced. Accountability mechanisms produce process more than accountability. The Treasury's bilateral structure blocks cross-cutting delivery.
They disagree on some of the remedies. They disagree on how big the state should be, on how much to centralise versus devolve, and on whether technology is the primary lever or a supplement. These are real disagreements, and Blueprint has its own views on them. But they don't undermine the core consensus. They're arguments about the best route to a destination that everyone agrees needs to be reached.
The question isn't whether to fix UK government delivery. It's how to fix it in a way that survives the next election. And that question — the question of institutional protection for reform — is where the next phase of the policy argument needs to focus. The diagnosis is done. The prescription is substantially agreed. The missing piece is the architecture that makes the cure stick.
Publication Index
All 101 publications referenced in this paper, organised by institution. Each entry includes title, date, and a brief summary.
Total: 101 publications. Compiled March 2026. Blueprint for a Better Britain is an independent policy initiative focused on institutional reform and government delivery. This paper is intended as a reference document for policymakers, officials, select committee staff, and researchers working on UK state capacity.