Sandra Naranjo Bautista is a public policy expert, specialising in governance, implementation, and public financial management. She is the former acting vice president, minister of planning and development, and minister of tourism in Ecuador, and has worked in Kenya, Indonesia, Mexico, and Papua New Guinea. She has worked as a governance specialist for OPM's ‘Learning for Action Across Health Systems’ project and as a public policy reform specialist with OPM in the QTAG project in Papua New Guinea.

Public policy is more an art than a science. Everyone recognises that scaling up is hard. There is an implicit acknowledgment of the challenges that organisations – including governments – face when they attempt to take successful pilots to scale; however, the translation of that understanding into concrete actions to overcome this problem is vague. Perhaps the real problem is truly embracing the implications of what ‘scaling-up’ implies – not just in terms of resources, but also in terms of strategy and systems. I would like to share three of the multiple lessons I have learned in over a decade working in, and researching topics about, the public sector.

1. Multiply vs modify

One of the simplest ways I have found to highlight the difficulties of scaling up is with a sweet analogy: baking a cake. After all, who doesn’t love cake? Imagine you are preparing a chocolate cake. You follow the recipe, make some variations of your own and it turns out to be delicious. Now you are asked to bake a cake three times bigger. To simply triple the ingredients is not enough; you also need to triple the strength used to mix those ingredients. At that point, perhaps your hand mixer might not be enough. Now imagine your cake becomes so famous that you are asked by your local store to bake a hundred of them. As you can imagine, now it is not only about increasing the amount of ingredients, but actually upgrading your capacity to produce cakes. A new mixer, an adequate oven, a team of people, and even a new kitchen might be required.

This simple example shows that sometimes multiplying, like in the case of the three cakes, is enough. There are governmental programmes in which multiplying successful pilots works out well. In the public sector that often occurs with complicated problems, like building roads. There are other occasions in which you need to modify your programme to be able to scale, like in our analogy with the hundred cakes. Malnutrition programmes are a classical example where successful pilots cannot simply be ‘scaled-up’. Stunting is a complex problem that requires a strong behavioral intervention and community engagement to succeed. While it is possible to deliver an intervention with a very high level of human contact and interaction in pilots, to scale it becomes virtually impossible. For a programme in child malnutrition to be successful, it will require a modification in the design of the intervention, not simply multiplying the resources to produce more of the same. Usually this will involve setting up systems ­– human, technological, and logistical, among others ­– that could sustain the intervention at scale. The art of policy design is then to be able to identify when to modify and when to multiply. These two concepts are often not differentiated.

2. Scaling up is not always possible or desirable

When a programme is working well, or there has been a successful pilot, there is always the temptation to want to go to scale. If a programme has been working well, why not? The problem is that scaling up is not always possible or desirable. Having a successful programme or pilot is necessary but not sufficient to go to scale.

Think of a building made with Lego. It looks beautiful as a miniature, but that doesn’t mean we could scale up that model and build one real size. It will need a different kind of material, scaffolding, and internal structure to withstand real-world conditions. In this case, scaling up is not possible. Now, imagine you were able to overcome these obstacles and it were possible to build it to real size, would it be practical? Probably not. It would not only be inefficient and costly, but very likely you would prefer not to live in a Lego building. In this example - a bit exaggerated to illustrate the concept - scaling up wouldn’t be desirable.

For programmes to be able to be implemented they need to simultaneously meet three key criteria, which is known as the strategic triangle. This framework, developed at Harvard University, refers to the need for programmes to be simultaneously:

• technically correct: when the theory of change from inputs to outputs and outcomes is adequate and has proven to work;  
• administratively feasible: having the financial, human, and logistical resources and systems required to execute, monitor, and evaluate a programme successfully; and
• politically supportable: having the required backing from the authorisers and relevant stakeholders to implement a programme.

Having a successful pilot or an effective programme does not guarantee that the conditions that made the programme initially successful will be met at scale. Frequently, there is the implicit assumption that because it worked before, it will work now. Often the analysis of the particular circumstances required to scale up, as well as the analysis of whether it will require modification or multiplication of resources, is rushed or omitted.

Going back to the child malnutrition example: while the logistics of an intensive human intervention are manageable at a pilot level, administratively it is exponentially more difficult to maintain such a ‘high-touch’ intervention at scale. While there could be political support to implement the programme and the theory of change could have proven to work, administratively such a resource-intensive intervention could end up being extremely costly. There is also no guarantee that the programme will still be effective when it is implemented outside of the carefully-controlled environment of a pilot or small-scale implementation. In that case scaling up the pilot may not be desirable or feasible.

3. Keep your eyes on the target

It might sound counterintuitive, but losing sight of the objective is more common than you might think. The day–to-day activities, along with the pressing need to show that you are doing something, make it easy to confuse the goal. Scaling up a particular project becomes the objective, rather than actually achieving the real results, which is why you started an intervention in the first place. The pilot or the programme were just a means - not the end in itself. In our example, our goal was to reduce child malnutrition and for that purpose a pilot, with a set of interventions, was designed. If at a certain point that final objective is not met, or it could be achieved in a more efficient way, then the scaled version of the pilot needs to be reassessed. The challenge is that it is always easier to mention how many children are attending antenatal check-ups and receiving nutrition supplements, or how much money the government is spending on the programme, rather than measuring the impact of the programme on the number of children who are malnourished.

If there is a successful pilot to offer nutritional supplements to children but scaling up is not working, you have one of two options. The first is to analyse what is going wrong with the scaling up the programme and try to fix the problem(s) to make the programme work. If the programme is still not working, even after you’ve tried multiple improvements, then perhaps there is no configuration of the pilot design that will work at scale. Instead, you may need to consider how systems can be introduced to facilitate the intervention at scale.

The second, less obvious, alternative - particularly if you have already invested money, time, and political capital in a project - is to stop trying to scale up the programme and consider an alternative, or even reconsider whether that intervention is suitable to scale up. At the end of the day, your objective is not simply to make a small programme large, but to reduce child malnutrition. What economists call a sunk cost can bias policymaking and induce us to make costly mistakes.

Scaling a successful pilot is just one example of why policy implementation is more of an art than a science. There is no magic formula to fix all the problems. A perfect pilot may not work at scale. In the end, it is about finding the right mix of tools for a policy intervention that would work for the particular context and circumstances that you are in. We can, however, learn from each other. There are three main takeaways when thinking about scaling up a programme or a pilot intervention:

• Identify clearly what is the objective you want to achieve and assess as many times as necessary the degree to which the scaled version of an intervention is helping you attain that objective. Always keep in mind the criteria of efficiency and effectiveness.

• Once you have identified a successful programme or pilot intervention, analyse it through the lenses of the strategic triangle to define whether it is possible and desirable to scale up the intervention. Consider the three elements of the strategic triangle for your evaluation and, even more important, consider what else would need to be done to ensure that these three conditions hold if the pilot were to scale.

• Finally, be sure to distinguish the type of scaling that would be required: a multiplication of the existing programme, or a modification that could require setting up systems – human, technological, and logistical, among others – that could sustain the intervention at scale.

A version of this article previously appeared at Imago.