‘What Works Centres’ are becoming an increasingly familiar part of the policy and professional landscape. In addition to the existing centres – covering more than a £250 billion expenditure – there is now a steady stream of interest in creating more.

Yet when the National Institute for Clinical Excellence (NICE) – in many ways the blueprint for later What Works Centres – was created in 1999, many doubted that it would survive to the next election. It was intended to get politicians, and the Department of Health, out of judgements about which treatments worked and for what. Yet senior figures feared that as soon as the young institution dared to say ‘no’ – that a given drug was not cost-effective – all hell would break loose.

Sure enough, it didn’t take long for angry calls to be received in Whitehall from a major pharmaceutical company demanding that NICE guidance against its product be revoked. Against the expectations of many, the line was held, and the first of today’s What Work Centres went on to help shape and guide the clinical practices of a generation.

At the time NICE was created, there was talk about creating a ‘social policy NICE’. Archie Cochrane himself nodded towards the need to extend such evidence-based approaches to other fields in an epilogue to his famous 1972 book Effectiveness and efficiency (Cochrane, 1972).

Yet it was not until after the 2010 election that serious work began into the wider application of the What Works approach and centres. A number of factors helped spark the decade of institution-building that followed.

First, the fiscal context sharpened minds. The UK faced a structural deficit of 8 per cent. Against this background, tough questions were being asked about what programmes and activities merited protecting – what worked and at what cost? Second, a weird part of No 10 – the Behavioural Insights Team – starting running randomised controlled trials.¹ The rapid and practical conclusions these gave rise to helped to popularise experimental methods. Third, senior figures in the administration, including the new Cabinet Secretary Jeremy Heywood, the Cabinet Office Minister Oliver Letwin and the Chief Secretary to the Treasury Danny Alexander, came to support the idea – not least with Jeremy’s decision to major on the idea at his first speech in 2012.

Alongside NICE, we borrowed heavily from the model and experience of Washington State Institute for Public Policy, and spent long sessions with its inspirational head, Steve Aos.² Indeed, it was an example from Washington State that persuaded UK Treasury ministers to first back the What Works approach – that Washington State legislature was persuaded against spending US$2 billion on a supermax prison on the basis that other youth interventions were much more cost-effective at reducing crime (just the kind of evidence the Treasury was looking for …).

By 2013, the first clutch of new What Works Centres had been founded. These covered education (the Education Endowment Foundation), early intervention (the Early Intervention Foundation), crime and policing (What Works Centre for Crime Reduction) and local economic growth (the What Works Centre for Local Economic Growth). Against the tough fiscal background, and the politics and fragmentation of Whitehall, the centres were very different in their funding and form, but they had a common purpose and ‘theory of change’. It was an evidence-based project built around a simple trilogy: ‘generate–translate–adopt’.

At the core of the project was ‘translation’ – to get useful evidence into the hands of frontline practitioners. Answers to the questions and decisions they had to make every day – and on the basis of the best evidence that could be assembled. As Steve Aos had put it, centres needed to produce ‘Consumer Reports’ – the US equivalent of ‘Which?’ – for busy public service professionals.

A decade on, there are now 14 officially recognised What Works Centres, many of which have written for this book.³ As I write, there are active discussions about creating new What Works Centres – or extending existing ones – into further areas such as climate change; food systems; gender and race equality; and other public service areas not currently covered (such as probation, prisons and other parts of the Criminal Justice System). There are also new challenges, such as addressing the overlap between the What Work Centres that touch on different aspects of youth, and of course squaring up to the weaknesses of the evidence base laid bare in many areas by the scrutiny of the What Works Centres.

Over the decade from 2011 to 2021, the What Works approach has moved from ‘guerrilla action’ to mainstream (or close to it). There have been many milestones along this journey. These include both breakthrough results by the What Works Centres, and procedural changes such as the progressive revision of the Treasury’s famous ‘Green Book’ to put more weight on the need for establishing ‘what works’.

One of the most important developments of recent years was, appropriately enough, rooted in an analysis of how much government was itself evaluating what works. An analysis conducted by Treasury, Cabinet Office and the Behavioural Insights Team found that only 8 per cent of more than £400 billion of new spending was subject to any meaningful or robust impact evaluation. It was a statistic that shocked ministers, and was highlighted publicly by Michael Gove in his Ditchley speech.⁴ It also helped catalyse growing Treasury interest, and was later picked up by the National Audit Office to add extra pressure.

As a consequence, in 2021, the Treasury funded the creation of the new Evaluation Task Force (ETF; jointly between the Cabinet Office and His Majesty’s Treasury). This incorporated and enlarged the original ‘What Works’ team in the Cabinet Office, but also added a substantial ‘Accelerator Fund’ for new evaluations; leverage over a renewed £200 million joint outcomes fund; and more muscular pressure on departments through the Spending Review to genuinely find out ‘what works’.

It is still early days to know the scale of impact that the ETF will have, but its influence can already be seen in the renewal of funding to existing What Works Centres and in the increased profile given to evaluation across government. In one of its earliest activities, the ETF organised a four-day event in Spring 2022 for public servants on evaluation, with speeches from Cabinet Office and Treasury ministers, sessions led by Permanent Secretaries from multiple departments; and of course a heavy presence from the What Works Centres.

We have come a long way from the uncertain early days of NICE more than 20 years ago. It is something to be celebrated. Yet we are still just scratching the surface. The UK spends around £1 trillion a year – itself a fraction of the global expenditure of governments. We have shocking little knowledge about which of this expenditure does good or harm, nor what alternatives might work better.

In the spirit of this book – as its authors seek to learn what works from ‘What Works’ – let me finally share my own suggestions on ten things we will need to do over the coming decade.⁵

1. 1.Map the gaps. What don’t we know? This should directly guide our research programmes and funders, including through the updating of department ‘Areas of Research Interest’, as well as likely spark a next generation of ‘What Works Programmes’ and where necessary the creation of new What Works Centres. At its core, we need to have the bravery and humility to be open about questions we don’t have the answers to.
2. 2.Build the skills. We need public servants – and a policy profession – that can tell the difference between a good and a bad piece of evidence, and has the skills and methods to build evaluation into programmes and systems. In a near future, it should be hard to get into the Senior Civil Service without demonstrating competency in evaluation and other methods that make it possible to answer the question: ‘What works?’
3. 3.Expand the pipeline. We need to increase the volume, as well as quality, of the What Works evidence base. This means innovation funds with robust evaluation; evaluation by default; and protocol publication (a early key performance indicator!).
4. 4.Build the data architecture. It’s currently just too hard to link and securely collate data on outcomes, intermediaries and interventions. In particular, we need to build up the capacity to rapid ability to assess multiple group or aggregate outcomes, with a data architecture or/or Application Programme Interfaces to make this possible.⁶
5. 5.Mine the variance. One paradoxical advantage of a long history of not identifying what works, is that our public (and private) services are awash with semi-random variations in practice. This creates a massive opportunity to ‘mine this variance’ in public services – through regression, machine learning and deep dives – to identify promising interventions. This variance is the ‘tropical rainforest’ of public service innovation – amazing discoveries of things that do, or don’t, work better, including for who, when and where.
6. 6.Reproduce. It’s hard enough to get things evaluated once, let alone going back and repeating the process. But it’s what we need to do, not only to increase our confidence intervals, but also to check on implementation – and to test further variations and improvements. We should borrow the adage of the skilled craftsperson: ‘measure twice, cut once’. We need to replicate, iterate, improve … and both check on implementation, but not fossilise practice.
7. 7.Increase the hit rate. Alongside increasing the pipeline of new interventions and evaluations, we can seek to increase our ‘hit rate’. Currently, typically fewer than one in four new interventions in social policy are found to be better than ‘business as usual’. It can be possible to increase this hit rate though careful prototyping (for example, to iron out issues pre-randomised controlled trial) and by identifying transferable solutions, techniques or insights.
8. 8.Translate and adopt+. Increasing the pipeline of evidence will have little impact unless matched by corollary increases in the volume and sophistication of our translational activities, and the absorptive capacity of those who the evidence is ‘aimed at’. So hand-in-hand with our plugging the gaps in the What Works landscape, we need to nurture evidence-curious professionals.
9. 9.Bring the public with us. Although public service professionals have been the prime early focus of the What Works Centres and movement, the public matter too. They are the patients, parents, consumers and voters whom are actually impacted. We should seek to nurture a nation of ‘bullshit detectors’ – who ask ‘Why do you think that’s the best thing to do?’ They are the ones in the room (Nsangi et al, 2017).⁷
10. 10.Go global. The UK, and a small clutch of other countries, are pressing ahead with building the What Works evidence. But the objects of education, public safety, economic growth and so on across countries overlap substantially. We should turn What Works into a global public good – through shared commissioning of systematic reviews, comparing contrasting approaches, and coordinating our research and development spend and institutional development through G20 and other international platforms.⁸

Imagine what’s possible if we deliver on even a clutch of these challenges or tasks. It would be transformative. A second enlightenment even.

The What Works Centres have already had considerable impact. But in many ways, what they’ve really done is shine a light on the depth of our ignorance. There are so many choices we face, from the billions spent on one (unevaluated) programme or another, to the decision as to what is the right or wrong thing to say to, or do for, a troubled youth that might set them on a better path.

I think, and hope, we will look back one day and puzzle on how we ever managed without them.