Test, Learn and Grow
At the Government Outcomes Lab, we investigate how governments can organise, collaborate and use evidence to improve outcomes for citizens. Our research explores how public systems can respond to complex social problems that require adaptation, experimentation and learning, while maintaining accountability for public resources.
The Cabinet Office-led Test, Learn and Grow (TLG) programme puts these questions into practice. As TLG’s learning and evaluation partner, we are examining how iterative, adaptative, and learning-oriented policymaking practices unfold in practice, under what conditions, and whether they lead to improved outcomes for the people and communities government serves.
What is Test, Learn, and Grow?
Test, Learn, and Grow represents a deliberate shift away from fully specified, top-down approaches to developing policy solutions. According to Nick Kimber, the Director of Public Service Reform (Place, Design and Innovation) at the Cabinet Office who leads the initiative, the idea for TLG was inspired by “the sense that a chronic failing in the UK is the gap between policy and delivery.”
The programme’s ambition is to close this gap by testing a way of working that emphasises local experimentation and feedback loops between different levels of government. Rather than assuming that a policy solution can be fully designed in advance, test-and-learn involves “starting small, learning quickly, and adapting based on evidence”. The aim is to generate real-time learning about implementation early enough for those designing and delivering public services to act on it and to improve continuously.
The Programme is working in ten local authority areas (‘places’) in England, each tackling a multifaceted policy challenge: from data and AI and Violence Against Women and Girls (VAWG), to child development, economic inactivity, neighbourhood health and special educational needs and disabilities (SEND).
In each place, an innovation squad, known as an ‘Accelerator’, brings together professionals from different backgrounds and levels of the system into a multidisciplinary ‘blended’ team. This includes, for example, professionals from central and local government, the voluntary, community and social enterprise sector, and disciplines spanning product development, technology and digital, service design and user research, system transformation and agile delivery.
The teams collaborate to scope a specific problem, surface the riskiest assumptions about how the system is currently tackling it, and test those assumptions through real-world experimentation. These ‘test-and-learn cycles’ are intended to generate timely and useful evidence about service delivery, human behaviour and system dynamics, that partners in place can use to adjust and refine their approach to policymaking.
TLG is not only concerned with what happens locally. The programme also asks what needs to change in the wider system if government is to act on what local experimentation reveals.
It’s second strand - ‘Grow’ - focuses on identifying and ‘growing’ reforms to central government processes and governance arrangements. The premise is that local teams’ ability to adapt can be shaped – and sometimes constrained – by centrally developed rules, processes and forms of governance. Without parallel shifts to its wider institutional environment, local experimentation will remain constrained by the system it is attempting to change.
This combination of local experimentation and system-level reform makes TLG particularly interesting for the GO Lab. Our research has long examined how institutional arrangements, relationships between organisations, and different approaches to evidence and accountability shape governments’ ability to improve social outcomes. TLG allows us to investigate these questions in the context of a live programme of public service reform.
GO Lab’s Role
As TLG’s independent learning and evaluation partner, GO Lab’s role is to generate robust, independent evidence about whether and how this model of reform works, under what conditions, and for whom. We are doing this by designing a developmental evaluation. This means balancing our proximity to delivery as evaluators embedded in a live TLG Partnership, while holding onto our analytical independence that keeps the evidence from evaluation credible.
What are we investigating?
To wrangle with the dynamism and complexity of a programme as large as TLG, our evaluation takes a multi-level approach and uses mixed methods to combine different strands of evidence, each suited to answering different kind of question. At the Accelerator level, we are exploring how test-and-learn approaches are adopted in practice across a diverse portfolio of policy areas and places, and where feasible, assessing their impact and value for investment. At the meso-level, we conduct comparative, cross-cutting analysis, looking at patterns in organisational routines and ways of working across all Accelerators over time. At the programme level, we probe how learning from Accelerators is synthesised, acted upon, and embedded within central government systems.
By evaluating both the ‘ways of working’ and the policy problems being tested, we hope to give policymakers, practitioners and researchers credible evidence on the conditions under which test-and-learn approaches can deliver meaningful, scalable improvements. At the same time, we hope to tease out the tensions such approaches create for accountability, assurance, and value for money in government.
