From education and mental health to early childhood, eldercare and more, many public service ideas that have been shown to work are not reaching everyone who could benefit from them. Numerous innovations are rigorously trialled each year, and many show promise. But effective models almost never make the leap to widespread impact. While there are pockets of excellence, lasting systemic change—at a scale that can improve population-level outcomes—remains elusive.

Why do innovations often shown to work in small-scale tests, then disappear? Why is it so difficult to bring effective new practices and programmes to greater numbers and diversity of people? And what might governments do to create the conditions to enable solutions that work to be scaled—so more people can benefit more quickly?

In short: what is meant by scaling, and why is it so hard?

‍

What, Exactly, is “Scaling”?

Scaling is different from testing a promising programme at more sites, and it is more than expansion or replication. It is an intentional process, defined in the literature as “maximising the reach and effectiveness of a range of actions, leading to sustained impact on outcomes”.¹ Scaling is achieved when reach is maximised alongside effectiveness—when everyone who could benefit from an innovation has experienced it, and when benefit is optimised for all.

This nuance is important because it brings two key challenges of scaling into sharp relief.

First, when considering reach, there is the centrally important challenge of equity. Without a focus on maximising reach to society’s most marginalised groups, scaling can inadvertently increase inequity. Studies show that when there is only partial spread of evidence-backed innovations, people who are already advantaged tend to benefit more than disadvantaged groups. The irony is that in spreading good practices and innovations we might, perversely, be entrenching inequality—when the intention is the opposite. For example, an intervention to ensure girls stay in school might work well in urban settings, but risks deepening rural-urban inequality if the delivery model is not suitable to reach children in remote communities.

Second, there is the importance of sustaining effectiveness when something is scaled. It is not enough for an intervention to reach everyone; it must also maintain quality and impact at scale. A common challenge is optimising the benefit of an innovation when it is introduced into complex social, economic, and political systems. As a simple example, efforts to scale a population-wide vaccination programme would be in vain if frequent electricity “brown-outs” mean refrigerated vaccines lose their protective benefits.

What Should be Scaled?

Clearly, we should only scale things that are shown to be effective. In practice, however, many innovations are scaled without robust evidence about whether or not they actually work.

Scaling without good evidence creates risks, the greatest one being that an innovation causes harm. A well-documented example of this is the programme Scared Straight, which aimed to deter young people from criminal activity by taking them on prison visits. This and similar programmes spread across the U.S. and other geographies, but rather than reducing offending, they actually increased the rate among young people.²

Even if an innovation is shown to be effective, scaling one thing necessarily involves trade-offs and opportunity costs. Leveraging resources for any scaling effort—whether public funds, healthcare providers’ time, or scarce political capital—means not investing in scaling something else. And it could (and often should) mean disinvesting in an existing service or approach.

So scaling is a choice, and one that needs to be carefully made.³

‍

‍

Even when good, well-evidenced choices are made about what and how to scale, the path remains challenging.

Innovations are Too Complex

New interventions often prove too complex to be adopted and sustained at scale.

An intervention may be too complicated for the system it operates in, is too expensive, or makes too many demands on participants. It might involve multiple components in proscriptive sequences or combinations. A model might only be deliverable by practitioners with advanced skills or qualifications, or require specialist roles that do not exist in many countries. Many interventions involve a complex architecture of training and re-training, certification and re-certification, supervision, and quality assurance—important for maintaining quality and “fidelity” to the original model, but a tall order for many hard-pressed public service systems.

Some innovations, particularly if tech-based, are beautifully simple and require only a mobile phone. These include mobile money services targeting unbanked people, apps for farmers allowing them to track crop prices and weather, or automated call services providing free parenting advice.

More complex interventions might have sufficient support when they are part of a well-managed evaluation or have government attention, but they often slip away when there are new priorities, or when key champions depart. An instructive example is the U.K.’s investment in a set of parenting programmes in the early 2000s.⁴ Despite a clear need for intervention, a proven model, and good early support, the programmes proved too demanding to scale effectively. Their presence waned significantly—as political support changed, and stakeholders moved on.

Simplification is key to optimising innovations for scale.

Simplification is key to optimising innovations for scale. Innovators should always aim to start as simple as possible: develop and test the least complex version of the model you think will work, and build up from there only if early testing suggests it is too simple.

Lack of Adaptation for Different Contexts

There is one truism about effective implementation at scale: “context is king”. Innovations must be actively adapted to local conditions. If an innovation is not well aligned with the particular environment it will operate in—the social conditions, the capacity of local delivery organisations, and the wider ecology of policy and funding—then it simply will not scale.

Not only do all these factors make scaling complex, but they are also all constantly changing: systems are not static. So innovations need to be adaptive, in a “dance” of alignment and re-alignment. Some innovations are unable to scale because they do not have adaptation baked in.

For an innovation to be scalable, we need to clearly understand its essential components—the “active ingredients”—that must stay unchanged to maintain effectiveness. Other optional or non-core elements can be adapted to suit different contexts.

Governments play an important role in supporting continued adaptation on the road to scale, with local leaders often best placed to support this work. One example of how local government played a key role is the scaling-up of Teaching at the Right Level in Côte d’Ivoire (known locally as Le Programme d’Enseignement Ciblé). This strongly evidence-backed approach was developed by Indian NGO Pratham and expanded to several African countries in a joint venture with Abdul Latif Jameel Poverty Action Lab (J-PAL).⁵ It was scaled through a donor-backed, multi-year programme of support that was government-led. When the Côte d’Ivoire programme ran into implementation challenges managing vital process monitoring data, local officials had the system context to address this.⁶ The core component—the availability and use of monitoring data—was maintained, with a different local delivery model.

“Voltage Drop”: Loss of Expected Benefits at Scale

Often, an innovation shows promise when tested in one or a few settings, only for impacts to dissipate or disappear when scaled up or applied in different contexts. This is known as “voltage drop”—the loss of expected benefits at scale. A recent analysis estimates this affects 50% to 90% of social innovations.⁷

Voltage drop often arises because the version of the innovation and its delivery tested is not the version scaled. Early tests are often done under “ideal scenario” conditions, which are not the same as when the innovation is promoted and delivered at scale in the real world. For instance, a health programme might have been tested in clinics that volunteer for research studies—with a culture of excellence and highly trained staff—whereas at-scale delivery would happen in all types of settings, not just the “best” ones.

Testing with scale in mind means doing so in real-world conditions: if we want to understand how an innovation will perform when scaled, we need to test the model under the conditions it will face at scale. John List, author of The Voltage Effect, gives an illustrative example: if we want to fly around the world, we cannot just test an aeroplane on a short journey from Chicago to Indianapolis.

We need to build scaling thinking and trialling into how programmes are built from the very start.

Innovators’ Capacity for Scaling

Innovators may be individuals, academic teams, start-ups, or groups working within NGOs or the public service system. Some have all the skills and capacity needed to take their innovation to scale. But many do not. They may be a small team, a hospital or health service with a local mandate, or an academic team invested in proving what can work rather
than scaling it up.

Innovators need to think right from the start about their endgame for scale. They might “spin out” from their current organisation, creating a start-up dedicated to scaling their innovation. They could set up a licensing system, to cascade an innovation widely. They could consider passing their innovation to a third party capable of funding and delivering it. They might partner with like-minded organisations. In some cases, the innovation may be so simple that it can be freely accessed online or purchased, with no support needed.

Many social innovations will aim for a path to scale through public service infrastructure. In fact, it has been said that “the road to scale runs through public systems”,⁸ as governments are uniquely placed to act as both delivery system and funder. It would usually be impossible for a social innovator or programme developer to establish a service delivery architecture as extensive as the existing public system. Thus, it is generally more effective and efficient to use the public system.

Each of these possible endgames might significantly influence models, partnerships, and data—so early thinking about the desired endgame is vital. Whatever the pathway, we encourage innovators to identify people and organisations critical to scaling their effort—and to engage them early.

For an innovation to be scalable, we need to clearly understand its essential components—the “active ingredients”—that must stay unchanged to maintain effectiveness.

Too Little Attention to Markets

In scaling commercial products, demand and market dynamics are important factors for success. But market viability is an oft-neglected concept when it comes to scaling social innovations.⁹

It is tempting to think “if we build it, they will come”, but scaling is never this easy. Understanding likely demand, as well as how an innovation fits into the existing marketplace of programmes and services, is as fundamental for social innovations as commercial ones. To scale, any innovation must secure its place in the market, through focus on factors such as competitive advantage, funding sources, pricing strategy, customer needs and buying behaviour, and target market segments. This is true even when the customer is the public sector.

Exploring such questions takes some innovators into very new territory. University enterprise zones, present in many countries, show—for example—how academics can be supported to advance real-world impact by bringing in commercial considerations, including industry and funder partnerships and capital to support testing and commercialisation.¹⁰

‍

‍

What can policymakers do to support scaling, in the face of these challenges?

Build Supply and Cultivate Demand for Evidence Use in Policy and Practice

The first principle for governments in choosing what to scale is to prioritise innovations backed by robust evidence.

Over the past couple of decades, governments have been investing in different approaches to incentivise or mandate the use of evidence in policy and practice. Countries such as the U.S. and Canada have established evidence “clearing houses”—online catalogues that identify proven approaches in specific service areas and provide summarised insights. More recently, in the U.K. and elsewhere, government-backed “what works” centres do the same work in compiling databases of evidence-backed interventions and sharing evidence about proven approaches in more accessible ways.¹¹ Integrated data systems enabling deep analysis of trends and needs (for example, leveraging de-identified tax, benefits, and health system data) can further support decision-making, either within public systems or managed within state-aligned, privately-funded institutions—such as the California Policy Lab or Research Improving People’s Lives in the U.S.

Governments can also mandate evidence use as a tool to spur evidence-informed decision-making, integrating this into their day-to-day business. For example, the Socio-Economic Impact Assessment System (SEIAS) in South Africa has, since 2015, required all policies that go before Cabinet to build an evidence-informed argument for policy options. Aotearoa New Zealand has created an independent expert function to advise the Prime Minister directly on how science can inform good decision-making.¹²

Initiatives like these are important aids to effective scaling, as they help ensure that policymakers focus on proven approaches when they look for what to scale.

Invest in Evidence and Skills for Scaling

There are many publicly funded evaluations under way for interventions that have no plans for—and no viable path to—eventual scale-up. This is a form of research waste.¹³ Scaling must be on the agenda of innovators.

Governments, their agencies and other funders of innovation and its testing, can play their part in asking tough questions of innovators that put scaling on the agenda from the start. Governments need to:

• encourage innovators to think about their endgame and work towards it from as early as possible, and
• encourage evaluations that replicate what would be the delivery and use conditions at scale.

Cultivate Institutions and Networks to Bridge Research and Practice

Scaling requires ongoing links and dialogue between the worlds of innovation, industry, evidence generation, service delivery, and policy-making. There needs to be an ecosystem of institutions and relationships to facilitate this work.

Governments can play a valuable role in cultivating platforms that bring these worlds together. For example, the Singapore government initiated a Families for Life Council as an intermediary to connect community service delivery partners and provide a platform to scale and sustain evidence-informed programmes that promote resilient families. A significant scaling effort was planned and initiated in partnership with corporate and public sector organisations over ten years. The initiative led to reductions in child behaviour problems and parental stress, and an increase in parental competency.¹⁴

City networks—such as the mayor-led Resilient Cities Network, or the Cities Idea Exchange, launched by Bloomberg Philanthropies to scale up innovations—can help government actors, programme developers, and funders learn from each other to advance projects and maximise reach and effectiveness.¹⁵

At a broader level, What Works Hub for Global Education, a new initiative led by the Blavatnik School of Government at the University of Oxford, brings together governments, NGOs, and academics to scale what works, enabling learning in low- and middle-income countries.¹⁶ The Hub is supported by the U.K.’s Foreign Commonwealth and Development Office, the World Bank, UNICEF, USAID, the Bill and Melinda Gates Foundation, and the British Council.

These and other examples illustrate ways in which governments are acting as connectors and brokers to develop the infrastructure needed to support scaling.

‍

‍

To reduce inequity and improve social conditions for those most in need, we urgently need to scale up impact. The use of good evidence should underpin government policy-making and resource planning, but it is also important to remember that scaling is a team sport.

Sometimes capital resources are needed, sometimes political agility, and sometimes well-nurtured talent. Policymakers can be most effective in supporting scaling if they consider the multiple actors that need to be involved, support the development of the connective infrastructure, and nurture the local champions necessary to lead scaling efforts.

We must get out of the habit of adopting things that do not work. And get into the habit of investing in the evidence, partnerships, and infrastructure needed to get what does work scaled.