Test and learn: a playbook for mission-driven government

The UK government has pledged to deliver bold new missions—ambitious, long-term commitments to tackle critical societal challenges. These promises will resonate strongly with many public servants who joined the civil service to change the world for the better. 

The question is how to deliver them. Missions try to address complex, wicked problems with tentacles that reach across government departments, local administrations, public services, industry, the academic community, and people’s personal lives. They are tricky puzzles where we can’t see all the pieces, and the structures and habits of the civil service are not well set up to take on cross-cutting challenges in an evolving environment.

Along with our colleagues at Nesta, we’ve thought a lot about what mission-driven government looks like. We’ve consistently advocated for a test-and-learn approach, which allows governments to pursue long-term mission goals while iterating and refining how they are achieved. It de-risks policy and funding decisions, helps us to navigate the uncertainties of complex adaptive systems, and supports the delivery of impactful, cost-effective solutions.

This evidence-based, iterative approach to government is not new, and there are teams within national and local government who use it to brilliant effect. The government’s Test, Learn, Grow programme announced by the Chancellor of the Duchy of Lancaster, Pat McFadden, in late 2024 is a fantastic initiative to nurture more agile and responsive forms of government, and a great investment in pragmatic, place-based problem solving in the UK’s public sector. 

The Test, Learn, Grow programme is part of a long-standing evolution in the way the UK government operates, which we’ve been supporting since our inception in 2010. When BIT was first set up within the British government, we embodied the test-and -learn approach: multidisciplinary teams working iteratively, in partnership, drawing on evidence and focused on outcomes. With partners across government, we delivered tangible improvements across a range of government priorities — increasing tax compliance through iterative testing of communications; boosting organ donation through user-focused changes to digital journeys; and increasing UK workplace pensions savings by billions of pounds annually by switching the default approach. This demonstrated the power of test-and-learn to deliver real-world change and impact at scale. But it’s far from universally adopted, and we still come across many public servants who feel that they don’t understand or are not equipped to take a test-and-learn approach to their work. 

We’ve just released a new playbook that aims to support government teams embarking on test-and-learn projects, whether as part of the Cabinet Office Test, Learn, Grow programme or independently. The playbook explains in detail how to build and run public policies and services using a test-and-learn approach, setting out a wide range of specific methodologies that practitioners can draw on at each stage of the innovation cycle. It’s intended as a practical resource for teams looking to design and test solutions in an outcomes-driven way.

In itself, this playbook is a reflection of iterative innovation in practice: the test-and-learn philosophy has been at the heart of the way that BIT has worked since we were set up in 2010.  In 2012, we published Test, Learn, Adapt, a paper that set out the case for iterative and evidence-based policymaking. We championed the randomised controlled trial as the gold standard through which policymakers could establish whether interventions were effective and how much difference they made, and we won the Institute for Government’s ‘Inspiration for Government’ award following its publication. We went on to publish TESTS, which explains the steps to follow when designing, evaluating and scaling projects in an iterative and responsive way.  We’ve now conducted over a thousand innovation projects in this fashion, and support partners around the world in developing the capabilities to run similar projects independently.

Our new playbook sets out more detailed considerations around embedding test and learn in government, along with a broader range of methods that can be used at different stages of the innovation cycle. These can be combined flexibly, depending on the stage of the policy or service cycle, the available resources, and the nature of the challenge – whether that’s improving services, testing creative new approaches, or navigating uncertainty in new policy areas.

Almost all of the methods set out can be augmented or accelerated by harnessing AI tools – from using AI agents to conduct large-scale qualitative research, to AI-enhanced evidence discovery and analysis, and AI-powered systems mapping and modelling. AI should be treated as a core component of the toolkit at each stage.  And the speed of evolution of the application of AI is another strong argument for maintaining an agile mindset and regularly updating our ways of working. 

We hope this playbook will make test-and-learn more tangible to people who are new to it, and will expand the toolkit of people who have more experience with the approach. And ultimately we hope it will serve as a practical cheatsheet for building and improving the fabric of life.

For more reading on the theme of mission-driven government and test and learn, check out:

• The recommendations from Nesta and the Institute for Government on how to set up missions, and the tools, systems, and cultures are needed for government to work in a mission-driven way. 
• The Catalytic State, by BIT’s President Emeritus, Professor David Halpern, which reviews the history of missions in government, and the approaches needed for success.
• The Radical How, from Nesta and Public Digital, which describes how to set up teams to make progress on complex societal missions.