Often radical social innovations come from seeing the opportunity to address a previously unsolved problem.

For many years, we thought that micro-credit was reaching people living in the harshest forms of extreme poverty. It wasn’t until researchers looked more closely in the 1990s that they found that in fact, it had failed to reach the poorest of the poor.

Enabling the poorest, most marginalized families to live better lives required not just tweaking an existing model; it needed a fundamentally different approach.

At BRAC, the international development organisation we work for, this realisation spurred a different model that led to the graduation approach, which aims to help people progressively “graduate” from extreme levels of deprivation. This strategy recognized that the experience of “ultra poverty” was not purely economic. We found that social support, community integration, and confidence-building seemed to change people’s aspirations as well as their income.

When we launched the graduation program in 2002, it was a new approach to tackling extreme poverty—in its selection process, the package of goods and services provided, the intensity of social support, and perhaps most significantly, the idea that after two years, participating house-holds could become independent.

What makes this model unique is its multifaceted approach. First, a targeting method is used, in which community members map out their village and identify the poorest people living amongst them. Those participants are closely vetted through surveys from BRAC. If selected, they receive services including a food stipend, financial literacy and skills training, healthcare, an asset like a goat or a cow to start their business, and weekly home visits by a BRAC staff member to ensure the participant is on track.

A randomized controlled trial by the London School of Economics found that five years after the program ended, 95% of participants in Bangladesh stayed on a positive economic trajectory. Last year, the model was rigorously evaluated in six countries around the world with similarly promising results. The research showed that impacts on participants from the program are endured and maintained one year after the program ends.

Now, our focus has expanded from implementing the program in Bangladesh to disseminating the model globally with graduation programs now active in over 40 countries.

But that doesn’t mean that we should forget about innovation.

Our experiences show that even as models become established and scaled up in certain contexts, innovation shouldn’t stop so much as evolve. Here are four areas that are relevant for many programs at scale.

1. Cut costs

Resources are always in limited supply – the more you can stretch them, the more people you can reach. In every program, there are usually opportunities to eliminate layers of the model to see if they are crucial or just nice to have – in the process reducing costs and complexity.

In our ultra-poor program, we wonder if households could receive a visit from staff every month instead of every week. What if food aid was only provided during the lean season? Though these adjustments may feel small at the household level, when you’re reaching thousands of households, these cost-savings can really add up.

2. Improve ease of adoption

To really take a model to scale, usually an innovation moves from just one implementer to many. Our ultra-poor program has now been replicated by a number of NGOs, governments and multi-laterals worldwide with different profiles and objectives.

The process of adopting a model into an organization is not always easy; many models that are effectively implemented by their creator fail when spread to others (for example, a successful oral rehydration therapy floundered outside of Bangladesh.) The best models are like furniture from IKEA – easy to assemble, whoever you are.

But packaging a model for alleviating ultra-poverty or mobilizing social change into step-by-step directions is hardly easy. We’ve realized for our program that a crucial tool for new implementers is a smart dashboard and e-learning modules that will enable them to collect the right information and put it to immediate use.

3. Increase applicability

Scale also usually requires a model to work in a range of contexts. For example, in Bangladesh, communities in the Northeast experience flooding for half the year, resulting in seasonal dependence on fishing and agriculture. Households in the south face changing soil salinity and weather patterns from climate change. Certain villages look to migration as a seasonal coping strategy.

Meanwhile, in the urban context, portable livelihood skills like mechanics or tailoring can lead to both entrepreneurship and employment. The model must be flexible enough to allow for diverse skills training approaches to ensure adequate livelihood opportunities. In some cases, the implementing organization can partner with government entities or other NGOs with this expertise if it isn’t available in-house.

4. Embrace opportunities to advance

Sometimes it’s the most successful models that run the greatest risk of getting left behind; because they work, we are reluctant to experiment with them too much. This can mean that we miss the opportunity to incorporate new technologies or trends into the model, even though these could increase impact beyond what the original design allowed.

For example, in many countries, including Bangladesh and Kenya, where mobile money has become a popular tool for financial transactions, the model could be expanded to incorporate digital transactions and financial education. This not only offers a secure way for people in the most rural, remote areas to receive and save their stipend, but offers a group with extremely limited access to financial services to master a powerful financial tool.

As the world takes a renewed commitment towards eradicating extreme poverty, it is important to note that social innovation remains a critical component of this scaling process. Maintaining the spirit of creativity offers opportunities for cheaper, simpler, and more adaptable solutions, with appropriate systems of support.

This article was originally published by the World Economic Forum. You can read the original article here.