The government’s project portfolio just improved. No project got better.

In March 2026 the National Audit Office told the Treasury something reasonable: the list of the UK’s largest, riskiest public projects had grown so long that it no longer told anyone much. Fix the categorisation, the NAO said, and put real governance around the handful of schemes that are genuinely too big to fail. A month later, the government did exactly that: it cut the Government Major Projects Portfolio (GMPP), the register every major UK public investment sits on, from 213 live projects to 81 overnight. Next year’s delivery confidence figures will almost certainly look better than this year’s. Not one of the 132 projects that left got any easier to deliver.

The reasonable case first

The NAO’s argument, made in its March 2025 report on major project governance, was not that the portfolio was too big for political comfort. It was that one register covering everything from Sizewell C to a regional case-management system could not give proportionate scrutiny to either end of that range. Mega-projects spanning multiple parliaments got the same quarterly review as programmes a tenth their size. The watchdog told NISTA and the Treasury to redraw the categories around risk and strategic importance, not just cost.

NISTA’s response, effective 1 April 2026, matched that recommendation closely. A project now has to clear three tests to stay on the GMPP: support a top government priority, carry a whole-life cost above £1 billion, and be judged to benefit materially from central scrutiny. A new “mega-project” tier sits above that, for schemes like Sizewell C, HS2 and the Dreadnought submarine programme. Chief Secretary to the Treasury James Murray called it directing “specialist expertise to the most complex, high-risk and strategically relevant projects.” Concentrating a stretched assurance function on 81 projects instead of 213 is a defensible response to a genuine governance problem.

What the reasonable case leaves out

The 132 projects that left the register did not disappear. Responsibility for their assurance “largely returns to departmental teams,” in the government’s own words, with central review only where Treasury approval is specifically required. Gone is the annual, published delivery confidence rating that could force a project’s problems into a document a select committee could summon a minister over. What replaces it is whatever governance the sponsoring department already had, for programmes that, by NISTA’s own account, were rated amber or red far more often than green.

That matters because of who is left holding the risk. The NAO’s own January 2025 report on government’s technology suppliers found that the public sector, which spends at least £14 billion a year on digital programmes, has repeatedly struggled to act as an “intelligent client”: scoping and overseeing large technology programmes without the specialist skills to catch problems early. NISTA calibrated the new £1 billion whole-life-cost threshold for physical infrastructure, where nine and ten-figure budgets are routine. Most government technology and service-delivery programmes, including some that have failed expensively, never reach that size. The capability gap the NAO diagnosed sits disproportionately in exactly the projects the new criteria now exclude.

A trick this portfolio has pulled before

None of this is new. The Institute for Government flagged the same mechanism in 2016, when the GMPP shrank from 188 projects to 143 and the share rated red or amber/red still rose to 32%, its highest level since 2013. That happened because a smaller portfolio meant the same handful of chronic problem projects, the Ministry of Justice’s shared services programme among them, counted for a bigger fraction of the total. A shrinking denominator moves the average even when nothing underneath it moves at all.

This time the contraction is far bigger. The 2026 cut removed 62% of the portfolio (132 of 213 projects), against 24% in 2015-16 (45 of 188): a reset two and a half times deeper, applied to a portfolio whose last published figures already showed red ratings at 18%, up from 12% two years earlier, and amber falling from 72% to 58%. Some of that shift is real: in 2025-26, 26 projects exited having successfully delivered, up from 14 the year before. But a portfolio completing more and failing more, then cut by 62%, produces no scorecard anyone can read as a clean trend.

The verdict

Both things are true at once, and leaders running their own portfolios should sit with that rather than resolve it. The NAO was right that 213 projects under one governance regime was unworkable, and the reset is a genuine fix for the 81 that remain. It is also true that the easiest way to make a portfolio’s headline numbers improve is to change what counts as the portfolio, and Whitehall has now done that twice in a decade, both times just before independent scrutiny was about to intensify. The European Court of Auditors reached a related finding in its January 2026 review of EU cross-border rail projects: average delays of 17 years and cost growth of 82% tracked with how many separate authorities held a piece of the delivery chain. The UK case is not cross-border, so the parallel is not exact, but the direction holds: assurance pushed outward is assurance that thins.

Anyone who runs a portfolio dashboard should take one thing from this. A rising green share is only informative if the scope it is measured against held still. Before crediting a portfolio’s improving numbers to better delivery, ask the question NISTA’s own report answers about itself: has performance actually changed, or has the count changed? In Whitehall’s case, for now, it is mostly the second.

Sources

• NISTA, Major Projects Annual Report 2025 to 2026, gov.uk (13 July 2026)
• GOV.UK news release, “Government refocuses major projects to boost delivery of national priorities” (31 March/1 April 2026)
• New Civil Engineer, “Government Major Projects Portfolio cut to 81 schemes as Nista aims to ‘target expert advice'” (31 March 2026)
• National Audit Office, Governance and decision-making on major projects (14 March 2025)
• National Audit Office, Government’s approach to technology suppliers: addressing the challenges (16 January 2025)
• Institute for Government, “How are government projects faring?”
• Institute for Government, “Government delivery of major infrastructure – less is more”
• European Court of Auditors, review of EU cross-border rail (TEN-T) megaprojects, reported January 2026