Whitehall Monitor 2026: What It Means for Digital

The Institute for Government’s Whitehall Monitor 2026 is essential reading for anyone interested in how government works. Now in its 13th year, the report provides an annual data-driven assessment of the UK civil service covering everything from workforce size and morale to pay, diversity, and departmental spending.

The report also covers digital transformation and artificial intelligence (AI), and for those of us working in this space, it offers a valuable stocktake of progress and challenges.

What struck me most was how this year’s report positions test-and-learn approaches, and specifically the Test, Learn and Grow programme, as central to addressing the challenges it identifies.

The digital transformation challenge

The Whitehall Monitor documents an ambitious agenda. In 2025, the Labour government set out its ambitions for ‘digital transformation’ to change how the state works. The government said the state should work “more like a start-up” and be “using the power of tech to modernise the system”.

Progress has been made. The Department for Science, Innovation, and Technology (DSIT), established in 2023, has become the “digital centre of government.” The Government Digital and Data profession has grown 152% since 2016. AI pilots are proliferating across departments.

But the report also highlights ongoing challenges. It points to what it calls a “culture gap” between policy and digital professionals. And on AI, it observed that while trials report time savings, there’s a need for more robust evidence on productivity gains.

These aren’t new concerns. Last year’s Whitehall Monitor 2025 noted that “there is little systematic way of evaluating pilots, or sharing what has been tried or deployed across government”. The 2026 report shows this is starting to change: and TLG is part of the reason why.

TLG as a model for evidence-led innovation

The Whitehall Monitor 2026 cites Test, Learn and Grow as one of the genuine bright spots:

“Pilots of the Test, Learn and Grow programme have provided proof of concept, demonstrating that new ways of working between Whitehall, local government and communities are possible.”

The report names TLG alongside the Government Communications Service’s ‘Assist’ AI tool as evidence that “with sustained, committed leadership government leaders do have the ability to change the way the system works”.

TLG isn’t just another pilot: it’s a £100m, three-year programme designed to make test-and-learn the default approach to policy and delivery.

Addressing the evidence gap

The 2025 report’s observation about the lack of systematic evaluation is precisely what TLG is designed to address.

The Government Outcomes Lab at Oxford’s Blavatnik School of Government has been embedded as TLG’s independent evaluation partner from the start. The ambition is to generate rigorous evidence about what works, for whom, and under what conditions.

The 2026 report notes that GCS Assist succeeded in part because the team “published two reports about what they had learnt”, and they were open about both what worked, and what didn’t. TLG is taking the same approach, with the Grow part of the programme specifically tasked with disseminating learnings from accelerators and the programme itself to other Local Authorities, and to central Government.

Breaking the cycle of contract lock-in

The Whitehall Monitor highlights a deeper structural problem: government’s reliance on legacy IT systems and third-party suppliers. The report notes that “around 80 legacy IT systems that the government were still actively using posed a critical level of cybersecurity risk”. It warns that in the rush to roll out AI, government “risks rolling out AI on to poor data held together by third-party suppliers, exacerbating government reliance on them at a time when government should be minimising it”.

This is where TLG’s approach offers something different. Rather than large programmes that go hand-in-hand with lengthy procurement cycles and long contract lock-ins, TLG encourages iteration in short cycles. The model is to test ideas quickly, learn from what happens, and adapt: building capability and evidence as you go, rather than committing to multi-year contracts before you know what works.

Crucially, TLG brings digital skills directly into its multidisciplinary teams, including DDaT professionals working alongside policy, delivery, and local government colleagues. We’re not outsourcing everything to large tech providers: we’re building the capability to do this work within government and with communities.

Bridging the policy-digital divide

The Whitehall Monitor 2026 recommends reducing what it calls the “policy-digital divide” through secondments and joint teams with shared accountability. This is exactly how TLG’s accelerator multi-disciplinary teams operate. The accelerators bring together people from central government, local authorities, and communities to work on shared problems.

And this is evidenced in the work so far. In Sheffield, the report notes, local government officials described a shift “from a hierarchical relationship with central government representatives to a more equal, collaborative partnership”.

The road ahead

The Whitehall Monitor 2026 concludes its assessment of TLG with an important observation:

“Test, Learn and Grow is itself a test of the government’s ambitions: outcome-focused, iterative and collaborative. The real test will come in seeing whether it can deliver on its promise to really change how the state works.”

The early results are promising. Wave 1 pilots drove measurable increases in family hub engagement. In Wave 2, we’re expanding to ten locations across a number of missions: family hubs, economic inactivity, neighbourhood health, violence against women and girls, and data driven front-line public services.

Comparing the 2025 and 2026 Whitehall Monitors, there’s a clear shift: from identifying the absence of systematic evaluation and lesson-sharing, to highlighting programmes like TLG that are building this infrastructure. That’s progress worth building on.