The Programme Constitution
Researched by an agentic pipeline · reviewed and gated by the author
The control objective is to preserve the enterprise’s ability to explain and defend the decision made when the boundary is crossed.
The visible reform is not the real reform
A designated UK mega project must now publish a short Strategy and Delivery Plan that states, among other things, its cost and schedule ranges, critical assumptions, stage gates, guardrails, change tolerances, descoping options, decision rights and reset conditions. The document is meant to be public, approved through departmental and ministerial routes, and updated when major decisions or a fundamental reset materially change the programme. [S1]
That is a striking list because it brings into the open the choices that large programmes usually postpone. What may be cut first? Which assumptions are load-bearing? Who may spend contingency? How far may cost, schedule, scope or benefits move before the programme must return for approval? The new framework requires these questions to be answered while the programme still has options, rather than after contractual, political and organisational commitments have hardened.
The reform should not be mistaken for the invention of reset governance. Stage gates, delegated tolerances, independent assurance, contingency and formal programme reset are established practices. UK government guidance published in 2024 already described reset as a structured intervention, with different reset types, phased work and decision checkpoints. [S6]
The deeper change is the attempt to combine familiar controls into a programme constitution: a compact that governs how uncertainty will be converted into decisions.
A baseline describes a plan; a constitution governs change
Most programme baselines answer a necessary but limited question: what do we currently expect to deliver, by when and for how much?
That question becomes less useful when the conditions supporting the answer begin to fail. A supplier strategy proves unworkable. A planning decision slips. Inflation invalidates a cost assumption. A dependency changes. The intended benefits remain valuable, but the original scope is no longer affordable. At that point, the decisive issue is not whether the programme is off baseline. It is whether the governance system can distinguish ordinary adaptation from a change that invalidates the investment decision.
A reset-ready programme therefore needs more than a baseline. It needs rules governing the right to change.
The UK framework makes those rules unusually explicit. Its guidance asks programmes to identify assumptions that would trigger reset or re-approval, define tolerances and escalation routes, prioritise descoping options, and state which decisions sit with the delivery body, programme board, department or ministers. [S1] The accompanying funding model adds broad early ranges, staged development commitments and, after final investment decision, a fixed capital envelope containing funded contingency. A forecast breach can lead to a fundamental reset decision on scope, funding and whether to proceed. [S3]
The control chain is therefore:
- uncertainty is acknowledged through ranges and assumptions;
- commitment is staged while evidence matures;
- tolerances define the space for delegated adaptation;
- pre-ranked scope choices make sacrifice deliberate;
- contingency ownership determines who can absorb which risks;
- reset triggers identify when the original authority is no longer sufficient;
- re-approval decides whether to change the scope, change the envelope or stop.
None of these elements is individually novel. The material shift lies in connecting them before delivery.
Why pre-committed sacrifice matters
Large programmes rarely fail because nobody can imagine reducing scope. They fail more subtly because the scope discussion begins too late, under pressure, when every component has acquired a sponsor, a contract, a dependency and a story about why it is indispensable.
Consider an illustrative transformation programme combining a new operating model, a data platform, regional deployment, supplier migration and workforce redesign. At approval, the board accepts a cost range rather than a point estimate. It also ranks the outcomes: regulatory continuity first, operating-model simplification second, advanced analytics third, regional customisation fourth. The board pre-authorises limited sequencing changes, reserves removal of regional customisation to the investment committee, and requires full re-approval if the data platform cannot support the minimum regulatory outcome within the funded envelope.
Eighteen months later, integration costs rise and the supplier migration slips. In an ordinary governance cycle, the programme may protect every promise by moving the completion date, drawing down loosely governed contingency and presenting a revised baseline after the choice has effectively been made.
In the reset-ready model, the choice is already structured. Management can resequence within tolerance. The board can release contingency for identified risks. The investment committee can remove lower-ranked customisation with a known effect on benefits. If the regulatory outcome itself is no longer achievable within the agreed envelope, the programme cannot disguise a new investment decision as routine change control.
The point is not that the earlier decisions will be comfortable. The point is that the organisation has preserved the ability to make them.
Pre-ranked descoping is especially important because scope is not a single quantity. Two reductions of equal cost can have radically different effects on outcomes, risk and future optionality. A programme constitution forces sponsors to decide which benefits are core, which are contingent and which may be traded away. That decision is more credible before every workstream becomes politically protected.
Reset-readiness is not the expectation that plans will fail. It is the decision to govern what happens when important assumptions do.
Ranges improve honesty only when they narrow
The use of ranges is another important but easily misunderstood feature.
Early point estimates often manufacture precision. A range can be more honest because it shows the uncertainty appropriate to the programme’s maturity. The Office for Value for Money model explicitly expects broad cost and schedule ranges to narrow as risks reduce and evidence improves. [S3]
But ranges can also become a shelter for weak control. A programme that remains inside a wide interval may still be deteriorating. A sponsor may describe movement towards the upper bound as anticipated uncertainty even when the underlying causes were avoidable. Accountability is not improved merely because variance has been renamed.
A useful range must therefore be connected to a theory of maturation:
- What evidence will reduce the range?
- At which stage gate should that evidence exist?
- Who independently tests the estimate?
- Which residual uncertainties are funded through contingency?
- What movement requires explanation even if it remains technically within tolerance?
- When does the range itself need re-approval?
The range is not the control. The control is the disciplined process by which uncertainty is reduced, allocated or escalated.
This distinction matters because the official reform is still a design hypothesis. The March 2026 guidance specifies the information that programmes must publish, but no completed cohort yet shows that the framework produces better decisions or delivery outcomes. The documents may improve transparency without changing behaviour. They may also legitimise drift if boards treat tolerances as entitlements rather than boundaries for judgement.
A reset declaration is not a deliverable reset
HS2 provides the strongest warning against confusing formal reset with restored control.
The National Audit Office reported in June 2026 that substantial work remained before the programme could complete its reset. A robust cost and schedule baseline was still needed, alongside commercial renegotiations, Euston plans and organisational capability. The programme was aiming to complete the reset in spring 2027, with potential cost reaching £102.7 billion. [S4]
The lesson is uncomfortable and useful: a reset is not completed when leaders announce that one has begun. Nor is it completed when governance is reorganised or a new date is selected. A deliverable reset requires credible evidence, executable commercial arrangements, capable leadership, explicit scope choices and a baseline that the organisation is prepared to govern.
This is why reset triggers should be designed before crisis. Under pressure, organisations have incentives to preserve momentum, defend prior decisions and avoid reopening the investment case. Pre-committed escalation cannot remove those incentives, but it raises the cost of evasion. It makes the decision that is being avoided visible.
The same caution applies to fixed envelopes. A fixed capital envelope sharpens the boundary between delegated delivery and a new political or investment decision only when the estimate is robust, contingency is genuinely funded and exceptions are transparent. Otherwise, the envelope becomes another number that can be revised after the fact.
The strongest sceptical case
The sceptical interpretation is that this is an elaborate transparency layer around ordinary programme controls.
That case has force. Mega projects are a narrow category. Existing business cases, assurance reviews, stage gates and change control already provide substantial governance. A public document cannot overcome poor estimates, weak commercial strategy, capability shortages or unwillingness to stop. The formalisation may add approval effort while leaving real decisions untouched.
There is also a scope problem. The Public Accounts Committee challenged the narrow definition of mega projects, noting that complex digital transformation, artificial intelligence and repeatable programmes may fall outside it despite their strategic importance. [S5] Meanwhile, NISTA reduced the centrally overseen portfolio from 189 projects at the March 2026 snapshot to approximately 80 from April, with other major projects managed through departments. [S2] A more focused centre may improve proportionality, but it also shifts greater assurance risk to organisations with uneven capability.
These objections sharpen rather than defeat the thesis.
They show why the reform should not be copied as a 20-page template or treated as proof that public disclosure improves delivery. Its value lies in a smaller set of governance disciplines: make assumptions explicit; stage commitment; define decision rights; rank scope sacrifice; fund contingency; and specify when the current authority expires.
For shorter, modular or reversible work, those disciplines should be lighter. A digital product that can release value incrementally does not need the same fixed-envelope logic as a nuclear or rail programme. A repeatable rollout may rely more on batch economics and stop-loss rules than on ministerial reset. Enterprise adoption should preserve the mechanism and discard the ceremony.
What boards should require before commitment hardens
A board does not need to predict every failure mode. It does need to know whether the programme can still make a clean decision when reality changes.
Before approving a long-duration, high-sunk-cost programme, the board should be able to answer:
- Which outcomes are non-negotiable, and which benefits may be traded?
- Which assumptions would invalidate the business case rather than merely change the plan?
- Which changes can management make without returning for approval?
- What evidence must exist at the next commitment point?
- Where is contingency held, what risks does it cover and who can release it?
- Which scope reductions are feasible, and what do they do to outcomes?
- What cost, schedule, benefit or dependency condition forces re-approval?
- Who has the authority to stop?
The final question is often the least operational. Many governance systems identify who may approve more funding but remain vague about who can withdraw authority. A credible programme constitution treats stopping, pausing and descoping as governed decisions, not admissions of personal failure.
The control objective is to preserve the enterprise’s ability to explain and defend the decision made when the boundary is crossed.
The test is behaviour at the boundary
The UK reforms make a serious move: they ask large programmes to disclose not only what they intend to deliver, but how they will govern the loss of certainty.
Whether that move improves outcomes remains unknown. Formal ranges can normalise drift. Reset triggers can be ignored. Descoping options can prove politically unusable. A smaller central portfolio can expose weak departmental assurance. The documents may become polished statements that follow decisions rather than constrain them.
The decisive evidence will appear at the boundary: when a major assumption fails, a forecast approaches the envelope, or a priority benefit is threatened. Does the programme use the authority it defined? Does it sacrifice lower-value scope? Does it reopen the investment decision before sunk cost becomes the argument for continuation?
A baseline tells us when the programme has changed. A programme constitution tells us whether the organisation still has the authority and courage to decide what that change means.
Sources
- National Infrastructure and Service Transformation Authority and HM Treasury — Strategy and Delivery Plan Guidance: Mega Projects — 26 March 2026 — https://www.gov.uk/government/publications/strategy-and-delivery-plan-guidance-mega-projects/strategy-and-delivery-plan-guidance-mega-projects
- National Infrastructure and Service Transformation Authority — NISTA Major Projects Annual Report 2025–26 — 13 July 2026 — https://www.gov.uk/government/publications/nista-major-projects-annual-report-2025-26/nista-major-projects-annual-report-2025-26
- Office for Value for Money and HM Treasury — Value for money study: governance and budgeting arrangements for mega projects — 19 June 2025, updated 24 April 2026 — https://www.gov.uk/government/publications/value-for-money-vfm-study-on-the-governance-and-budgeting-arrangements-for-mega-projects/value-for-money-study-governance-and-budgeting-arrangements-for-mega-projects
- National Audit Office — High Speed Two reset — 29 June 2026 — https://www.nao.org.uk/reports/high-speed-two-reset/
- House of Commons Committee of Public Accounts — Governance and decision-making on major projects — 10 September 2025 — https://publications.parliament.uk/pa/cm5901/cmselect/cmpubacc/642/report.html
- Infrastructure and Projects Authority — Resetting Major Programmes — 4 December 2024 — https://www.gov.uk/government/publications/resetting-major-programmes/ipa-resetting-major-programmes-html