When taxpayers entrust the state with their money, they want to know it will be spent wisely. They expect the government to set out its mandate and priorities, decide how it will invest to deliver and then ensure that the core department spending – of over £650 billion per year and rising – does not go to waste.
Today, more and more citizens feel shortchanged by this process, as the costs of government rise and outcomes worsen. This feeds into the great unmet demand in politics as voters, no matter their political leanings, become pessimistic and impatient about politicians’ ability to deliver tangible results.
In the UK, the central mechanism for (re)allocating public money is the multi-year Spending Review, when government departments and the Treasury negotiate their spending plans for the next two-to-three years. Based largely on bilateral negotiations between each department and the Treasury, and shaped by internal power dynamics, this is a crucial mechanism to align the entire machinery of government around a common purpose and a shared set of trade-offs.
As the chief secretary to the Treasury observed in March 2025, it is common to imagine the Treasury acting like a finance department, with different parts of a business reporting up to it in real time. In practice, however, it operates more like a bank, agreeing to provide money under certain conditions and then collecting limited information on how it is spent.
This is very different to how successful private-sector organisations manage their money, with entire portfolios of projects or assets managed in real time. Data are joined up, and dedicated systems for analysing, tracking and forecasting performance and risks routinely enable agile, timely decision-making.
The current approach to the Spending Review is so bureaucratic and labour-intensive that almost any technology-powered improvements would save time and money. Under this Labour government, the Treasury has begun to address some of these shortcomings. Yet the potential impact of its reforms is being badly undermined by deeper structural problems with the process itself. Commentators have noted that the latest cycle has fallen short of early promises of a technically transformed, silo-busting and truly “mission-driven” Spending Review.[_]
Fixing the Spending Review has to be about more than delivering efficiencies. A comprehensive redesign of the system itself is needed, putting in place an AI-powered real-time portfolio-management system. This would enable a periodic “Strategic Review” to replace the current clunky process, focusing on a refresh of government objectives and alignment of programmes to common goals.
This would amount to a radical overhaul of the Spending Review as we know it – and would represent a genuine step-change from passive budget management to active portfolio stewardship. Public spending should be managed like a strategic investment, with risk and return expectations for different programme types effectively addressed on appropriate timeframes.
Doing so would ensure that the centre of government has greater grip of, and agency over, the strategy around public spending. Blunt spending controls can be replaced by genuine accountability and transparency, removing a major source of political and operational friction between ministers and No 10, the Treasury and the Cabinet Office.
The best programmes would get funded, both at large portfolio-rebalancing points and as part of a regular routine of quarterly reviews. Initial spending bids can be drafted jointly by officials from different departments using AI co-pilots to share global best practice, surface lessons from similar previous bids and enforce minimum standards of evidence. Submitted proposals can be automatically analysed for alignment with government priorities, scored against strategic outcomes and merged with similar bids. Instead of duplicated or siloed efforts, departments can work together on cross-cutting problems.
The delivery of successful bids can be tracked against agreed milestones across all of government. Emerging evidence of outcomes can be captured and analysed, and financial performance monitored, so that failing projects are stopped or turned around while successes are scaled quickly.
Where data predict rising demand, funding can be surged strategically to support proactive governance, not crisis management. Waste can be prevented before it accumulates. High-performing departments and programmes that embrace the new approach could be given more authority and agency to deliver free from onerous oversight.
Success can be rewarded, and failure will have nowhere to hide.
The technology to build this kind of system already exists. With the government’s first Spending Review complete, now is the time to rethink how the next should run. Action today will lay the foundations for a different kind of process in two years’ time. This should be done rapidly and iteratively to quickly test, learn and develop. The government should:
Rapidly pilot the new approach with a single department, quickly building a minimum-viable version of a tool that integrates financial and performance data into a shared dashboard. This would help the Treasury to understand spending on key programmes and the department to manage it.
Publish lessons learned from this exercise, including its impact on outcomes, and use them to inform a full-scale rollout across departments, rewarding early adopters with greater autonomy on spending.
Build a secure web-based collaborative bid-drafting and submission tool, and test it during upcoming fiscal events.
Over time, build in more advanced AI and data capabilities, including advice on bid strength, an evidence explorer and duplication detection, working closely with users to rapidly iterate and evolve the system.
In time for the start of the next Spending Review cycle, develop a tool that integrates spending and performance metrics from across government, and is available to departments as well as the Treasury, the Cabinet Office and the prime minister. This should be used in regular stocktakes and Cabinet meetings, so that by the time of the next Spending Review the government is able to use the tools to operate a new kind of Strategic Review focused on outcomes and alignment to overarching objectives.
The UK’s public finances deserve a system fit for the 21st century, and the next Spending Review must not be another missed chance. For the Treasury, it is an opportunity to improve the quality and impact of public spending. For No 10, it is an opportunity to grip government and reshape it to be more focused, more dynamic and higher agency. And for government, it is an opportunity to earn back the trust of the people it serves.
Chapter 1
Public trust in government has reached historic lows,[_] with citizens increasingly questioning whether their tax money is being spent wisely. Recent research by the Tony Blair Institute for Global Change (TBI), titled Disruptive Delivery: Meeting the Unmet Demand in Politics, highlights this crisis as part of a broader trend: a collapse in confidence, not in democratic ideals or specific ideologies. Voters across the political spectrum demand results – on the cost of living, health care and public safety – but perceive governments as falling short. Core to that confidence is the state’s most essential function: managing public money effectively.
The UK government’s Spending Review (SR) is a central mechanism for managing public money. It allocates over 2.2 trillion[_] in predictable public spending and tries to align departmental budgets with policy priorities. These multi-year reviews determine expenditure limits for departments – resource departmental expenditure limits (RDEL) are set over three years while capital departmental expenditure limits (CDEL) are set over four years – and block grants for the devolved administrations.
The modern SR process was introduced in 1998, with then Prime Minister Tony Blair calling it a “new, strategic approach to public spending”.[_] It was designed to move beyond short-termism by creating a more coherent framework – one that maintains fiscal sustainability, aligns spending with ministerial priorities and ensures value for money by funding what works.
Today, the SR represents one of the most important and most time-consuming exercises in UK government, costing – at a minimum – tens of millions in civil-service time.[_] It dominates the life of senior officials and ministers alike for months on end, drawing their attention away from most other matters, and results in decisions that are difficult to reverse until the next SR cycle. In a single parliamentary term, a government would typically only go through one or two SR cycles. Technocratic as the process might seem, getting it right is therefore central to any government’s delivery ambitions.
The Current Spending Review Process
While the specifics of each review vary according to political and economic context, the process follows a broadly consistent and largely manual structure:
Strategic framing and departmental commissioning. In parallel, the Treasury commissions updated economic forecasts from the Office for Budget Responsibility. These projections – alongside internal modelling – determine the extent to which the government is on track to meet its fiscal rules and, crucially, how much fiscal headroom is available. This assessment shapes the overall financial envelope available for public spending over the review period.
Departmental submissions and iterative review. Departments then submit their spending bids, using lengthy narrative business cases and spreadsheets. Treasury spending teams – often generalists with limited subject-matter expertise – then manually review hundreds of documents in compressed timeframes. An iterative “challenge process” follows, where departments are pushed to refine, justify and in some cases reconfigure their bids to ensure alignment with policy intent and value for money.
Negotiation and settlement. Treasury teams manually prepare detailed analysis and strategic options for ministers. These briefings underpin often convoluted negotiations – often led by the chief secretary to the Treasury – with departmental leads. Final decisions are codified in settlement letters from the chancellor, setting out agreed funding and performance expectations for each department.
From that point on, the process shifts to performance management until the next SR cycle begins. Treasury oversight is largely confined to the OSCAR II (Online System for Central Accounting and Reporting) monthly return, when departments upload spreadsheets that track cashflows at an aggregate level but offer little insight into delivery progress or achieved outcomes. This retrospective, financial lens leaves departments with full visibility and control over how funds are spent, while the Treasury’s line of sight diminishes.
The result is a loss of continuity: the strategic intent behind funding decisions becomes decoupled from how spending is monitored and managed in practice. What begins as an effort to allocate resources towards outcomes reverts quickly to a system preoccupied with narrow budget tracking rather than financial and performance management.
The SR was introduced not simply as a tool for managing public money, but for governing with purpose. It is meant to reflect the government’s values, priorities and appetite for reform. And when used effectively, it can help drive long-term improvements in outcomes, efficiency and public trust.
Yet in practice, the largely analogue way in which government operates means the SR in its current form can fall short of its potential to align efforts across government, offering neither the clear strategic direction nor the rigorous, evidence-based public finance management and value-for-money assurance it is designed to provide.
Even with basic digital tools – like spreadsheets and form documents – helping to process submissions, the process still relies heavily on subjective judgement and personal influence over transparent assessment or standardised evaluation. As bids are developed, information remains heavily fragmented across emails, documents and disconnected systems, often opaque not only to outsiders but to the civil servants and politicians involved.
At its worst excesses, the process sees departmental civil servants on the receiving end of vague instructions from Treasury officials with insufficient guidance on how to structure a bid or business case to meet requirements. Departments often have a shortage of key skills around finance, analysis, evaluation and delivery, giving heads of finance and other operational officials within departments considerable and sometimes arbitrary influence over what gets submitted.
When ministers prioritise submissions, they consider a range of reasons including personal and political preferences in a context where the system cannot give them a clear idea of the impacts and trade-offs. Bids that lack any evaluation or basic feasibility may slip through, while well-developed ideas are quietly culled for bureaucratic reasons.
The government has initiated a series of structural reforms to address some of the long-standing weaknesses in the SR process. Most notably, it committed to a regular two-year review cycle, with each review setting resource budgets for at least three years – a change now enshrined in the Charter for Budget Responsibility. Alongside this, the government will introduce a ten-year infrastructure strategy,[_] providing greater stability and longer time horizons for investments that require sustained planning and delivery.
At the outset of the SR process, the government announced its intention to make departments collaborate through mission clusters, organising their bids around the government’s five cross-cutting missions. This approach was intended to address some of the limitations of bilateral SR negotiations by fostering cross-departmental collaboration and encouraging bids that reflect shared goals to support more strategic alignment. However, the results of SR 2025 show the missions approach failed to meaningfully supplant entrenched patterns of departmental bargaining.[_]
A further reform was the creation of the Office for Value for Money (OVfM) – a dedicated time-limited Treasury unit tasked with improving efficiency and strengthening oversight. OVfM worked directly with departments during the SR process with the intention of scrutinising investment proposals and rooting out waste. Between this and the next SR, it will lead thematic value-for-money reviews, targeting high-risk or cross-cutting areas of spending to ensure that the government maintains accountability for delivery over time.
In January 2025, the chief secretary to the Treasury also launched a set of reforms aimed at modernising the SR process itself,[_] centred around the introduction of shared digital dashboards. These were designed to provide information on costs, outputs and impacts across departments, and act as a single source of truth between departments and the Treasury on spending plans and the overall envelope.
Alongside this, the Treasury began using an internal large language model, nicknamed HMT-GPT, to analyse departmental submissions and generate cross-cutting insights. HMT-GPT can help surface thematic links between proposals, identify geographical and policy overlaps, and help connect spending plans to intended outcomes.
These changes mark a genuine shift towards greater transparency and a more evidence-informed approach to public spending. By enabling shared visibility and cross-departmental analysis, they can begin to tackle long-standing frustrations around opacity, duplication and inefficiency of the current process. They represent a significant step forward in equipping the Treasury with a clearer view across departmental bids and a more systematic basis for scrutiny.
However, while important, these changes remain partial. Without end-to-end digitalisation, their impact will be blunted by fragmented data systems, ongoing reliance on manual processes and fragmented communication, inconsistent evaluation standards and a heavy administrative burden on civil servants that limits operational efficiency. More fundamentally, these reforms stop short of addressing the deeper structural problems of the SR: entrenched information asymmetries, misaligned incentives and missed opportunities for strategic coordination, adversarial dynamics between departments and the Treasury, and the lack of continuous oversight and accountability once funds are allocated. Real transformation will require a comprehensive redesign of the system itself.
Information Asymmetries Create Perverse Incentives
Chief among the structural issues that continue to undermine the effectiveness of the SR process is the persistent information asymmetry between the Treasury and departments at different stages, which skews incentives and weakens collaboration.
Power dynamics in the SR favour the Treasury. At the start of the SR process, Treasury ministers and officials have more information and so have more power. In particular, Treasury officials have superior information about fiscal space and cross-government priorities. This creates the risk that some may influence decisions in favour of their preferred programmes (or preferred departmental colleagues). Within the Treasury itself, each spending team focuses only on one department, leaving them with limited understanding of what other departments may be requesting and so with limited visibility into likely settlement ranges. While this process strengthens the Treasury’s hand, it also means duplication and siloed thinking remain built into the process. Opportunities to spot duplicated projects and opportunities for silo-busting efficiencies are inevitably missed.
Black-box dynamics create perverse incentives to collaboration. Given departments do not know what others are submitting, the process creates a sort of “prisoner’s dilemma” – where entities acting in isolation may make suboptimal choices because they cannot coordinate with one another – that can encourage overinflated bids in some cases and unfeasible underestimates in others. Treasury officials, often pressed to make cuts, will attempt to identify areas for real cost-cutting opportunities while departments attempt to distract them with red herrings in order to reduce the risk of losing any potential budget. This kind of zero-sum cat-and-mouse game undermines transparency, collaboration and trust. As with all spending decisions, one department’s success means losses somewhere else in the system. In the absence of shared information about the trade-offs involved, this becomes, in a meaningful way, not dissimilar to a game of poker: the winners may not be the ones with the best proposals for how to spend money well, but those who are most successful at bluffing and who know exactly when to fold.
The process suffers from institutional memory loss. High churn among civil servants and the multi-year nature of the SRs means few understand the full process or remember previous failed bids and programmes. Record-keeping is sparse, at best inaccessible, making human memory the only available source of information. Incentives do not exist for individuals within the system to collect, curate and share information that could drive better outcomes or help learn from past failures.
Transparency fears lead to risk aversion. With transparency comes accountability. Making data and information about how SR decisions are made and the outcomes of those decisions more widely available would undoubtedly benefit the taxpayer and enable government to drastically improve how it makes decisions. It would also lay bare any systematic and some historic failings – which may appear threatening to any officials who might suspect that they were underperforming.
Time pressure undermines innovation. Although the process itself runs for months, the real SR timetable ends up highly compressed towards the end, cramming complex decisions inside the Treasury and tense negotiations with departments into a small number of weeks running up to the final settlement. This leaves limited capacity for process innovation, discouraging officials from trying anything new.
The Spending Review Is Time-Consuming and Carries Large Opportunity Costs
The SR remains among the most expensive processes in government when measured by civil-servant time. In particular, senior leadership across government is greatly distracted for months during bid preparation as their continued funding relies on success in this process. As continued programme funding is contingent on SR outcomes, in many significant ways departmental action is also often frozen until funding certainty is restored, affecting everything from hiring to purchasing and delivery.
In addition to the considerable staff time the review cycle consumes, the SR’s partial digitalisation – its reliance on spreadsheets and discrete documents – creates a deeper strategic shortfall. Without a consolidated evidence base, a cross-government view of duplication or effective stop-checks for underperforming spending, the process fails to support the difficult choices that effective strategy requires. This leads to sub-optimal trade-offs that undermine the SR’s core purpose of clear-sighted prioritisation:
Inadequate use of evaluation. At this point in time only approximately 34 per cent[_] of government major projects are evaluated effectively and 27 per cent are not evaluated at all – despite Treasury guidance requiring all bids to include evaluation plans. This marks a radical improvement since 2019 when only 8 per cent were properly evaluated, with 64 per cent having no evaluation of any sort.[_] This is largely attributable to the creation and effective implementation of the Evaluation Taskforce, a team led and staffed by expert evaluation specialists and ministerially empowered to make changes. While evaluation itself does not directly drive positive outcomes, it can be used to assess and review whether spending is effective, and to give important clues as to whether to invest larger budgets or kill programmes that are not delivering the desired outcomes.
Duplication across departments. The siloed SR process invariably funds very similar programmes in multiple departments, with insufficient systematic attempts made to find efficiencies or complementarity between programmes. Spending teams work in isolation, each engaging with departments individually rather than considering proposals across government as a whole. Without massive structural changes or effective oversight to locate and address inefficiencies, this will not change.
Inability to stop ineffective spending. Once funding is granted, it typically continues without ongoing systematic review. While the OVfM aims to provide new capability to tackle waste, there are still no systematic feedback loops that evaluate spending effectiveness against intended outcomes. Without mechanisms to stop or redirect underperforming programmes, wasteful delivery often continues unchecked, and futile delivery of ineffective policies and projects rolls on, with jobs secure indefinitely. The Treasury is known to prefer to claw back funds from transparently failing programmes and so civil servants are not incentivised to acknowledge failure. However departments rarely face consistent consequences for poor performance, and there is little benefit to succeeding when even strong programmes struggle to achieve funding to scale in the SR melee.
In addition to the opportunity cost of time spent on SR negotiations, there is a broader issue: the “opportunity cost” of approving one bid over another, which is rarely given sufficient attention in the SR process,[_] particularly when it comes to trade-offs between departments. This is a mistake. When financial institutions retrospectively assess their performance, they carefully track the cost of having spent money on one investment over another that would have generated higher returns. Strategic use of performance data to learn to reduce opportunity cost, alongside timely decisions to halt spending on programmes that are not generating expected returns and divert it to more profitable ones, are at the core of effective portfolio management.
Governmental decision-making around the SR fails to effectively weigh the impacts of sub-optimal investment across the full range of programmes, for the most part considering each department’s proposals separately. Failure to directly address and measure the opportunity cost of previous SR decisions means opportunities to learn and improve future rounds are missed.
Weak Oversight Undermines Value for Money
The lack of transparency, evaluation and guidance for effective spending bids significantly undermines the strategic purposes of the SR. After funding is allocated, however, the information asymmetries reverse. Departments now know exactly what they have been allocated and have (close to) full control over implementation, while the Treasury loses visibility of spending outcomes and delivery progress. Without comprehensive evaluation data systematically linked to funding decisions, departments continue funding ineffective programmes. There are few means to assess if outcomes that were expected at the SR stage are being achieved and limited opportunities for course correction.
While the chief secretary’s announced reforms represent a significant step forward in improving the process, systematic portfolio-level assessment of value for money across and between departments remains limited:
Despite improved data harvesting, bilateral negotiations still discourage optimal cross-government resource allocation.
The cycle of information asymmetry continues to create systemic waste through gaming and padding bids.
While transparency is improving within the Treasury, limits to broader transparency reduce accountability for delivery.
The power dynamics, perverse incentives and costs of the Spending Review process suggest it requires radical reform. This can be achieved through a complete digital transformation that combines transparency with better guidance and, consequently, a greater degree of autonomy for departments to deliver on agreed outcomes.
Chapter 2
Rather than incremental improvements to a model rooted in the technical limitations of the 2010s, government should reimagine the core process around the new capabilities of today’s technology. This means reshaping how government conceptualises, allocates and manages its financial resources, and associated timescales.
Instead of a periodic, opaque and labour-intensive exercise, the SR should become a continuous, transparent and data-driven process that optimises resource allocation on an ongoing basis with longer time horizons.
This is necessary because while current reforms may improve visibility, they are yet to address the fundamental challenge of the existing process: there is little active management of the entire portfolio of programmes and projects funded through the SR, which are rarely revisited until the next cycle.
This is in stark contrast to the operations of successful businesses in the financial-services industry. These businesses generally operate state-of-the-art data-management systems, working from a strategic position in which risks and potential rewards are balanced and hedged for security, and in which the mantra “cut your losses early and let your profits run” is ubiquitous.
Government, by contrast – and despite overseeing vast public budgets – generally fails to cut its losses in a timely way or ramp up investment in successful programmes, which falls short of its financial duties to the public purse. To a great extent, this is because the process of allocating funds and monitoring spending outcomes is both inefficient and ineffective.
A system that supports the development of higher-quality, evidence-based bids, maps all proposals against strategic objectives for prioritisation and de-duplication, tracks delivery in close to real time and ultimately embraces a risk-based portfolio-management approach across government is needed. This will help prevent costly failures, scale up successes, and deliver better services and more effective reforms at a lower cost to the taxpayer.
An Improved Bid-Submission Process
A comprehensive digital overhaul of the bid-submission process would start by replacing all remaining narrative documents and spreadsheets for the final submission with standardised digital submission forms. These would use uniform data structures and mandatory fields to ensure consistent information across all bids. In line with modern cloud office suites such as Google Docs, it should be possible for several people to work on a submission in parallel, including officials from different departments collaborating on a bid. This is currently challenging as departmental document-management systems are not interconnected.
Within the bid-design system, an integrated AI system can analyse proposals as they are being drafted and provide advice to civil servants, some of whom might lack business backgrounds, for example on how to build a robust business case, identifying gaps and suggesting improvements.
Critically, the system can also be used to enforce minimum evidence standards for all submissions. Proposals that lack adequate evaluation plans or relevant supporting data can be flagged, focusing Treasury scrutiny where it would have most impact and reducing the ability of civil servants to leverage relationships to bypass checks and balances. An evidence-explorer database can be built alongside to help explore best UK and global practice or examples of similar efforts failing in the past.
Data captured through this process and retained in a repository would eliminate remaining fragmentation of information, creating a single authoritative source for all bid information and supporting evidence. This would enable powerful cross-cutting analyses extending far beyond current Treasury or departmental capabilities, allowing comprehensive assessment of which bids succeeded or failed and why, and later comparison with outcomes. The system would maintain comprehensive records of all submission trajectories, including which applications are cancelled, by whom and for what reasons, creating real accountability and learning opportunities. For successful bids, this information can also be used to provide context for future monitoring efforts, allowing for informed comparison between expectations and reality.
Mock-up of a screen depicting improved bid submission
Source: TBI
Note: The images are provided for purely illustrative purposes to showcase what the proposed system with integrated AI capabilities, based on technology available today, could look like. The illustrations and text within them should not be taken to represent a production-grade IT system, data about live or proposed government programmes, or actual analysis of their potential costs or benefits.
An Advanced Comparative Framework
To maximise the strategic impact of spending decisions at the bid-assessment stage, proposals should be systematically mapped against government priority outcomes. This would involve extending current dashboard capabilities to create comprehensive visibility of how proposed spending connects to stated strategic objectives. Building on existing cross-cutting analysis, this would enable full cross-departmental visibility of related proposals, revealing opportunities for collaboration and highlighting potential duplication before funding decisions are made.
The existing AI capabilities should be enhanced with algorithms trained to identify similar initiatives across departmental boundaries and suggest consolidation opportunities. Standardised evaluation frameworks tied directly to strategic objectives would replace remaining inconsistent assessment approaches, ensuring that all proposals are judged against consistent criteria aligned with government priorities. This standardisation would further reduce the influence of negotiating skill or personal relationships on funding outcomes.
Mock-up of a screen depicting an advanced comparative framework
Source: TBI
Note: The images are provided for purely illustrative purposes to showcase what the proposed system with integrated AI capabilities, based on technology available today, could look like. The illustrations and text within them should not be taken to represent a production-grade IT system, data about live or proposed government programmes, or actual analysis of their potential costs or benefits.
Comprehensive Delivery Tracking
In the aftermath of the SR, comprehensive delivery tracking would fundamentally alter the post-allocation information asymmetry, create greater transparency and enforce accountability. Funding allocations would be explicitly linked to delivery milestones, creating transparency around implementation progress. Enhanced dashboards – as close to real-time as possible – would build on current systems to provide both the Treasury and departments with comprehensive shared visibility of spending against outcomes, replacing remaining information gaps with continuous mutual awareness.
While the OVfM could help to reduce waste through periodic thematic reviews for each major programme, this system would enable continuous monitoring on different timescales appropriate to different initiatives, supporting more timely interventions and allowing cost-saving strategies to be employed proactively. Over time, this type of ongoing technical delivery tracking would remove the need for oversight responsibilities to be vested in newly created offices or units. Instead, accountability would be built into the system by default, saving money and streamlining the process. This model makes something like the OVfM ultimately redundant once operating at full scale. Early warning indicators would automatically flag at-risk programmes based on delivery data, enabling proactive intervention before problems escalate. Continuous evaluation feedback loops can ensure that evidence on effectiveness continuously accumulates and informs ongoing resource-allocation decisions rather than being gathered sporadically, if at all.
Mock-up of a screen depicting comprehensive delivery tracking
Source: TBI
Note: The images are provided for purely illustrative purposes to showcase what the proposed system with integrated AI capabilities, based on technology available today, could look like. The illustrations and text within them should not be taken to represent a production-grade IT system, data about live or proposed government programmes, or actual analysis of their potential costs or benefits.
Full Portfolio Management
Taken together, these changes would lay the foundation for government spending to be reconceptualised not as a collection of individual departmental budgets but as a strategic investment portfolio with varying risk profiles and return expectations. Spending categories would be explicitly classified according to risk level (high/medium/low), with appropriate oversight and flexibility scaled accordingly.
This approach would introduce systematic mechanisms to reallocate resources from underperforming programmes to more effective alternatives, based on continuous evaluation data. A cabinet-level portfolio-review process would provide political oversight of this reallocation, ensuring that technical assessments remain aligned with democratic priorities. The system would enable rapid response to changing circumstances while protecting core spending, providing the agility that government increasingly requires in an unpredictable world.
The biggest cultural shift lies in moving from the current periodic-review approach to active management. The Treasury should introduce a quarterly “programme-cancellation exercise” – using real-time data to review funded projects and stop those failing to deliver. Funding could then be redeployed to higher-performing or emergent needs. Departments that participate effectively could retain a portion of redeployed funds, incentivising responsible risk-taking and proactive failure management.
Critically, this approach recognises that not all failure is bad. A government trying new things should expect some initiatives to fall short; a failure to fail at all is a failure of ambition. The goal is not to eliminate failure but to fail consciously, faster and cheaper.
Mock-up of a screen depicting full portfolio management
Source: TBI
Note: The images are provided for purely illustrative purposes to showcase what the proposed system with integrated AI capabilities, based on technology available today, could look like. The illustrations and text within them should not be taken to represent a production-grade IT system, data about live or proposed government programmes, or actual analysis of their potential costs or benefits.
A New Model for Managing Public Spending
Building directly on existing and planned new initiatives, including modern digital dashboards, AI tools and the OVfM, this comprehensive transformation of the SR would include these essential components:
Enhanced structured digital forms, which would replace all remaining unstructured and legacy-style documents, capturing standardised data that enable systematic analysis rather than idiosyncratic interpretation.
Advanced duplication-detection capabilities, which would extend current AI harvesting to leverage natural language processing and machine learning to identify conceptually similar proposals across departmental boundaries, even when described using different terminology. Comprehensive interactive dashboards would build on existing systems and data, and be designed to collect new data, to provide appropriately timed visibility into both allocation decisions and subsequent delivery progress, creating unprecedented transparency for decision-makers at all levels.
Advanced portfolio analytics, which would introduce concepts from investment management into government resource allocation, enabling risk profiling and return-on-investment metrics across the spending portfolio. Different spending categories could be assessed according to appropriate criteria – with different expectations and tolerances for innovative, experimental programmes compared with established service delivery.
Comprehensive digital-governance workflows, which would capture approval decisions and their rationales transparently, creating an auditable trail of decision-making currently lacking in the largely verbal negotiation process. These workflows would incorporate appropriate separation of duties and approval hierarchies while maintaining the speed and responsiveness required for effective government operation.
Together, these enhanced features would complete the transformation of the SR from a periodic, opaque and labour-intensive exercise into a continuous, transparent and data-driven process that optimises resource allocation across government. Active spending management would reduce the need for a periodic large-scale exercise focused on making the numbers add up.
Instead, government could use this time to set out its priorities and an overall spending envelope which meets its fiscal rules, ensure the current portfolio of programmes and new bids align to these priorities and envelope, and then set forward-spending limits. A shared foundation of data and AI analytics – akin to the National Policy Twin proposed by TBI last year in Governing in the Age of AI: A New Model to Transform the State – would help avoid the SR setting strategy by default, with No 10 working closely with the Treasury to run different scenarios, assess trade-offs and set out the right “portfolio allocation” for the whole of government. Far more than a process tweak, this would in fact constitute a radical change in how public spending is managed – a new model that replaces periodic Spending Reviews with a new “Strategic Review” approach.
Chapter 3
The chief secretary’s reforms have begun to address a critical tension currently preoccupying government finance leaders: balancing the long-term nature of much government activity and alignment to core strategic objectives with the need for decision-making that is agile and responsive to a volatile, rapidly evolving environment.
Many stakeholders have long been advocating for longer SR periods – typically five years or more[_] – to provide greater predictability for infrastructure projects, R&D funding and digital-transformation programmes that inherently span multiple years. This recognises the inefficiency of forcing long-term initiatives into short-term funding cycles. The current chief secretary has responded with the announcement[_] of a ten-year infrastructure strategy and ten-year R&D budgets[_] alongside Phase Two of the current SR to address “chaotic and short-term”[_] infrastructure delivery. The new methodology aims to set long-term capital budgets on a two-year cycle but is specific to infrastructure builds and does not include digital transformation, which remains dangerously under-resourced and poorly delivered.
Simply extending SR timeframes also introduces its own challenges, potentially hampering government’s ability to respond to emerging threats, changing population needs or shifting political priorities. The events of recent years – from the pandemic to geopolitical instability – have demonstrated the need for fiscal agility alongside predictable funding. Some programme timescales should be made shorter rather than longer.
Completing the portfolio-management approach begun through current reforms would resolve this apparent contradiction by providing differentiated treatment for different types of spending. Core infrastructure, R&D and transformation programmes would receive the long-term funding predictability they require for effective planning and delivery. Meanwhile, a portion of departmental budgets would remain flexible for reallocation as circumstances evolve and evidence accumulates on programme effectiveness.
This portfolio model would enable government to accelerate successful programmes mid-cycle without waiting for the next formal SR, while simultaneously providing mechanisms to adapt to evolving priorities without undermining the certainty needed for long-term investments. An approach which recognises that not all government spending should be managed according to identical rules and timeframes is needed.
This would enable responsible, close-to-real-time oversight over whole-of-government spending. Government would be able to monitor whether programmes are directly driving the outcomes and strategic objectives it sets out in speeches and manifestos, and improving the services they are mandated to deliver.
Ongoing active management and rebalancing would move the Treasury into a new space; rather than holder of the purse, it would be responsible for driving the system to deliver provable, well-thought-out outcomes. A multi-year Spending Review in its current form would no longer be needed, and the major fiscal event would now focus specifically on prioritisation – a Strategic Review to rebalance investments in line with the priorities of the day.
Fully Realised Evidence-Based Decision-Making
Fully digitalising the SR and embedding AI tools would finally realise the elevation of evidence in spending decisions. Current aspirations for evidence-based policymaking still collide with practical limitations in data availability, consistency and application as well as civil-service culture. Complete digital transformation would help create self-reinforcing dynamics to address remaining limitations.
An enhanced platform would extend current capabilities to automate rejection of proposals that fail to meet minimum evidence standards (either existing evidence or plans to gather evidence as appropriate), creating a powerful incentive for departments to invest in proper evaluation. This enforced discipline would progressively raise the quality of evidence throughout government, replacing the current uneven landscape in which evaluation quality varies dramatically across and even within departments.
Consistent data standards across submissions would enable meaningful comparison between proposals in different domains, supporting more rational prioritisation decisions. The system would accumulate evaluation data over time, building an increasingly valuable evidence base to inform future allocations. Enhanced machine-learning algorithms could identify patterns of success and failure across similar programmes, generating insights impossible to discern through traditional analysis of siloed evaluations.
Most importantly, comprehensive delivery tracking would extend beyond current dashboard capabilities to create a tight feedback loop between spending and outcomes, ensuring that evidence of effectiveness (or a lack thereof) promptly influences resource-allocation decisions rather than being discovered years later during the next review cycle.
Maximised Accountability and Transparency
A fully digitalised Spending Review would dramatically enhance accountability throughout government resource allocation and utilisation.
For the first time, there would be a clear line of sight from initial allocation through to eventual outcomes, enabling stakeholders to track the complete lifecycle of public spending. The system could provide comprehensive documentation of decision criteria applied to funding allocations, eliminating the opacity that currently shrouds these consequential choices. Comprehensive real-time visibility of delivery progress would eliminate information asymmetries between the Treasury and departments, creating shared awareness of implementation challenges and achievements.
Advanced automated flagging of off-track spending would ensure that problems cannot remain hidden until they reach crisis proportions, as frequently occurs even in existing digitalised systems. Early identification of issues enables corrective action before substantial resources are wasted on ineffective approaches. This transparency would fundamentally alter the dynamics between the Treasury and departments, replacing adversarial negotiation with collaborative problem-solving based on shared information.
The grand bargain here is one of embracing transparency in exchange for greater autonomy. Departments or programmes with a proven track record of good decision-making processes, financial management and delivering effective outcomes could earn a lighter-touch audit process from the Treasury and the Cabinet Office, with greater attention paid to persistently troublesome departments and programmes. This would help shift from a low-trust, low-agency system to one based on higher trust and higher agency to deliver the outcomes citizens need.
As part of this, government should invest in the analytical capabilities of departments, and create shared performance and financial data sets across No 10, the Treasury and government departments. The comprehensive transparency could extend beyond government itself, potentially enabling appropriate parliamentary and public scrutiny of spending patterns and outcomes. This democratic accountability represents the ultimate safeguard against misallocation of public resources, complementing the technical improvements in process efficiency.
Maximised Efficiency Gains
A fully digitalised process would deliver comprehensive efficiency gains throughout the resource-allocation system. The administrative burden of bid preparation and assessment would be further reduced through comprehensive standardised forms, automated validation and consistent evaluation frameworks. This would free senior leaders throughout government to focus on strategic priorities rather than document production.
The systematic elimination of duplicate efforts across departments would generate substantial savings, as would the comprehensive identification of underperforming spending. Most significantly, the ability to reallocate resources more agilely to high-value initiatives would increase the overall return on public spending, delivering better outcomes within existing fiscal envelopes. These efficiency gains would compound over time as the evidence base strengthens and the culture adapts to the new approach. The long-term potential is nothing less than a fundamental improvement in the productivity of government spending – an imperative given the fiscal constraints and service-delivery challenges facing the UK.
Chapter 4
An Accelerated Test-and-Learn Approach
Given the strong start of improvements already announced by the chief secretary and the critical importance of the SR process to government operations, complete digital transformation should now proceed at pace, while maintaining appropriate caution and building on proven success. The aim should be to have the full system in place, with experience of using its constituent components across government, by the time of the next SR.
To proceed at this pace, we recommend an accelerated test-and-learn approach that builds on current progress while mitigating risk and maintaining momentum toward the desired end state.
1. Pilot the approach with a department. The Treasury should rapidly prototype a system to integrate a single department’s finance and performance data into a real-time shared dashboard, generating AI-powered insights that will allow officials to understand departmental spending on key programmes as well as enable better management within the department. This would enable the Treasury to track spending in real time as needed – allowing it to significantly loosen existing “spend controls” requiring paperwork and meetings to sign off already budgeted large departmental spending – saving time and frustration on both sides.
This system should be built in-house by government, building on existing experience in small technical and engineering teams such as GDS (the Government Digital Service), 10DS (10 Downing Street Data Science) and i.AI (the Incubator for Artificial Intelligence) rather than as a tendered consultancy process, and should be completed within six months. Additional engineering resource should be onboarded into government systems where needed, and the system built with and for real users, following best practice in user-centric design and phased delivery.
2. Create an accelerated rollout plan to expand learnings to other departments by 2028. Learnings from the pilot will be useful in assessing the technical capabilities available and required to deliver a fully digitalised system more widely. Working with the Evaluation Taskforce, the impacts of this pilot programme on outcomes and behaviours should be assessed, and outcomes published for accountability and wider learning. Should delivery prove successful, the team should be expanded at this point and dock in with GDS to create a rollout plan designed to deliver the solution in a staged way across the whole of government.
The starting point for the accelerated rollout should be departments with strong financial-management systems, a track record of successfully adopting innovation, and a high willingness to share data with No 10 and the Treasury to create a single source of truth on spending and performance. As an incentive to encourage departments to participate, No 10, the Treasury and the Cabinet Office should relax Treasury-delegated authority limits and Cabinet Office process controls for high-performing departments.
3. Build a new digital platform for drafting spending bids, developed rapidly by an in-house software-development team. A simple secure web-based service by which departmental teams can collaborate on bid development and then submit their proposals should be achievable within weeks. To overcome the “blank page” problem, it should be possible for users to upload existing Word documents or spreadsheets to pre-fill fields in the form. Centralising and rationalising data about how bids are built and pass through various stages of approval will in itself be useful in understanding the system, including uncovering biases and vagaries in bid-approval processes. This system should be in place in time for the next minor fiscal event.
4. Build the Treasury’s AI capabilities to improve its evaluation systems. Next, the system should be advanced by building in AI capabilities to assist teams in writing more effective, detailed, evidence-based bids, and to help Treasury staff understand the strength of the evidence base. These tools should offer a shared view between the Treasury and departments, and should:
Offer feedback on the bid strength to the submitter.
Assist the bidder to ensure that the correct information is available and presented to the spending team.
Enable both the bidders and spending teams to uncover whether other bids, either in progress or submitted, are likely to contain overlapping deliveries, and enable bids to be connected to identify possible efficiency gains.
Develop a model to understand what makes a bid more likely to be successful in order to address systematic biases.
Identify deliverables, milestones and metrics automatically, and set up a process to track these. This offers the chance to identify whether programmes are falling off-track earlier, without manual intervention.
5. Build a single interface to view spending across all departments. Finally, an interface should be developed that will enable the Treasury to view spending across all departments, visualising information about delivery-cycle lengths, major milestones, crossovers between projects that may provide opportunities for more efficient delivery, delivery tracking of whether programmes are on or off target, and flags to indicate when data are sparse or missing. It should be possible to use this system to run stocktakes of all major (and eventually all) government spending, quickly identify where programmes are exceeding expectations, where greater investment is likely to drive better outcomes, and where projects are failing to deliver on time and in a cost-effective way, enabling closer to real-time portfolio management of major spending.
Each of these capabilities will require several test-learn-iterate cycles to develop and solidify the new system before the next SR. Starting immediately allows for thorough development and testing of enhanced capabilities before the next major SR exercise. All components of the enhanced system should continue to be tested on smaller fiscal events, such as annual budget adjustments or targeted spending rounds. These expanded applications would validate the approach while containing any implementation risks.
Delivery Considerations
New systems must be professionally designed and competently managed, requiring greater expertise in technical delivery within the civil service itself. The cost of such a team, with salaries set (in line with recommendations from the AI Opportunities Action Plan) at a benchmark of no less than 75 per cent of equivalent private-sector pay, is minimal compared to the overall cost of the process. Senior responsible owners (the individuals accountable for a programme or project meeting its objectives) should be effectively and transparently held to account.
The poor delivery of OSCAR II, an unambitious and innovation-free programme launched in 2021 to collect key management and accounting information in an aggregate form, underscores the importance of staged and rigorous delivery. OSCAR II struggled due to capacity constraints, poor initial design, a lack of effective and appropriate testing, and issues with access, all of which have had cascading effects on the UK government’s ability to produce timely and accurate financial reporting. These shortcomings not only hampered financial oversight but also undermined public trust. Every effort must be made to ensure that new systems do not repeat these mistakes.
Initially, the new systems proposed here should run alongside current improved methods, allowing for comparison and validation of results. This redundancy provides assurance against system failures while building confidence in the new approach.
This accelerated approach builds on the current momentum while balancing the urgent need for completion against the prudent management of implementation risks. By demonstrating significant wins on small-scale fiscal events, it should be in a good position to build confidence gradually and overcome natural resistance to change in critical government processes before the next SR.
The chief secretary’s digital reforms have begun a crucial transformation, but the prize of fully modernised government spending management remains within reach. Building on recent foundational work of partial automation and greater data transparency, the next phase must complete what has been started: introduce a new model of smarter public spending.
The approaching 2027 Spending Review represents a historic opportunity to transform how government stewards taxpayers’ money. The current partial improvements have demonstrated both the potential and the necessity of comprehensive change. Completing this transformation would not merely enhance government efficiency – it would restore public trust by proving that government can adapt, improve and deliver results in the digital age.
The choice is clear: accelerate toward full digitalisation now to enable a new type of Strategic Review, or accept that half-measures will continue to fail the public and undermine the delivery of critical public services.