In depth: How NGOs must harness data for good

Bridging the data divide requires NGOs to be intentional about how they collect and use data

So, how do NGOs practically work through differences in access to technology, education, and resources that affect communities’ abilities to collect and analyze data. How do they effectively “bridge” the data divide?

It starts with being intentional and investing in innovative solutions. NGOs are in a unique position for leveraging their current efforts to harness and utilize data more effectively through:

1. Collection: Recognizing the value of a “boots-on-the ground” approach to data collection
2. Communication: Measuring and communicating data effectively
3. Creation: Developing data-driven support solutions, potentially including AI applications

Collection of data

Data is the foundational element and the necessary ingredient to measure impact, communicate the value of interventions, and to construct creative solutions.

But before any of that can be realized, data collection – of course done ethically and responsibly – is needed. For Orbis, an NGO focused on eye care, the process involves surveying eye health within communities. Their data collection includes inputs such as retina photographs, visual acuity, and adherence rates to therapy. For others, it may be obtaining economic or environmental data points from underserved communities. In each case, acquisition of data rests on an approach perfectly suited to NGOs – being physically present in communities and being actively involved with the groups and people served.

NGOs can also address the problem of data diversity by collecting data from populations who have otherwise been ignored. There is increasing awareness that data is not inherently objective, and without applying an equity lens to the collection process, subsequent uses of that data – whether with communication or creation – will reflect certain biases. This notion surfaces often, for example, in conversations around algorithms developed without the inclusion of data from underrepresented groups. By contrast, NGOs are uniquely positioned to tackle this issue head-on by providing data from populations in which they are immersed and from which data is not typically collected; this leads to more representative data and models.

Developing large datasets can seem daunting to NGOs, but often, they are already doing more than they realize. Looking at where they are already working, the populations they are already serving, and the data that’s already being collected programmatically can be a great first step in the data-collection journey.

To take an example: Orbis collects data on refractive errors in children in India in a large school-based vision-screening program. This was part of their routine data collection, but over time, they eventually collected data from more than 2 million children in over 10,000 schools at multiple locations across India. An effort that began for programmatic reporting led to creating the largest ever dataset on child eye health. Using this data, Orbis published research, the first of multiple planned papers, to document the prevalence, severity, and determinants of refractive errors among school-going children in India, and beyond.

Sticking solely to a quantitative definition of data, however, is one way that NGOs limit themselves. What Orbis found equally helpful was to think holistically and broaden the concept of data to include not just quantitative data, such as medical images and exam findings, but also qualitative data, such as questionnaire responses and association of outcomes. This approach bolstered the organization’s ability to tell meaningful stories with the data – for example, by layering cultural context atop numerical findings.

Communication of Data:

Collecting data is just the first step. When NGOs start to look at the volume and depth of the data they have already generated, consolidation and analysis can lead to valuable findings. The most obvious advantage is to measure the impact of interventions, but NGOs should not stop there. By going on to communicate those findings internally or externally to donors, governments, beneficiaries, and other key stakeholders, they have the potential not just to improve programs and target interventions more equitably, but also to help advance sector approaches or advocate for change with decision- and policymakers.

Previously limiting themselves to basic demographic data for donor reporting, Orbis now places an emphasis on capturing impact through outcome-based data. In Orbis’s case, conducting an analysis and then undergoing peer-reviewed publication of existing program data proved to be a turning point – it acted as an "a-ha" moment for their teams on the ground, triggering a profound shift in their mindset regarding data collection. Publication as a goal can help inform data collection. The demands of the peer-review process – including institutional review boards, requirements for high-quality and traceable data capture, and most importantly, an understanding of why each data element collected is relevant – accelerate the time from data collection to data communication.

Spreading the reach of an NGO’s communication is important, and for that, we’d recommend that organizations look beyond their mission. For example, Orbis is focused on eye-care training, but the organization looks for cross-sector outcomes in areas such as education and economics. Concrete examples include the study from India previously mentioned, which found that near-sightedness appears to increase as literacy rates do, thus linking outcomes around vision to education. Other cross-sector examples include a recent peer-reviewed publication on the impact of vision loss on traffic safety among bus drivers, and one on the economic impact of near-vision loss on women garment workers. Communicating impact across sectors helps to broaden an NGO’s appeal and strengthens the viability and importance of its programs and work.

Strong data can also support influential advocacy work, increasing the lasting impact of an organization’s work. Orbis found expanding relationships and linking their work beyond eye care is again beneficial and demonstrates how their work contributes to larger issues. A valuable exercise for Orbis has been linking their work to the United Nations Sustainable Development Goals, beyond goal 3, which is health-related, to others, such as education, poverty elimination, and gender equality. This shift has opened doors to a broader understanding of how data can be utilized beyond reporting, enabling Orbis to assess and communicate the true impact of their work.

Creation:

We’ve outlined how effectively collecting and communicating data supports program planning and demonstrates the evidence-based value of the organization’s work. The next logical step is to create systems with the data. NGOs can use a variety of data-driven support solutions. Orbis has embraced implementing AI solutions, both with their own in-house algorithms, Cybersight AI, and through partnerships with AI companies, and we think other organizations can take advantage of the ways AI is becoming more accessible, too.

With that being said, NGOs are vastly different in purpose, scope, and data maturity, so each should proceed in alignment with its own needs, but the overall process and the questions NGOs need to ask themselves when developing a new solution are the same:

1) What problem are you trying to solve?

At Orbis, they knew that patients who were diagnosed with diabetic retinopathy would often not show up to follow-up appointments. They wanted to find a way to increase adherence to follow-up care; through AI, they’ve been able to improve access to medical diagnosis and referral uptake.

Another NGO, Reading Partners, focused on providing reading support to under-resourced schools, using AI to proactively identify locations with lower-performing scores and take action to improve performance.

2) Where will the solution fit into the workflow?

Integrating new solutions into existing workflows can improve efficiency and accuracy, but it requires careful planning and implementation. In Orbis’s case, they wanted as minimal friction as possible, so they designed the AI system to work in the background, on top of processes that were already familiar, such as a camera system linked to a computer capturing an image of the patient’s eye.

3) How will you efficiently develop a solution?

While working with technical partners may be necessary, there are an increasing number of low- or no-code solutions that NGOs can use (Leif Sundberg, Jonny Holmström, Democratizing artificial intelligence: How no-code AI can leverage machine learning operations, Business Horizons, 2023). NGOs may find that there are academic or for-profit systems that are interested in the data and with whom partnerships could be explored to ease the burden of augmenting various data sets. Given that NGOs have the “oil,” they can find ways to partner with other organizations in ways that are not extractive to their beneficiaries, but rather improve and expand the services NGOs provide.

4) How will you implement the solution?

To deploy a new solution, NGOs must not only integrate it into their workflows, but also provide training and support to users. NGOs should recognize their systems as iterative, requiring testing and maintenance to uncover potential bias or blind spots.

Orbis used the framework above to develop Cybersight AI, which can detect abnormalities often associated with common eye diseases – like glaucoma, diabetic retinopathy, and macular disease – in mere seconds by analyzing images of the back of the eye taken during routine examinations. When diseases like these are caught and treated early, patients have the best chance of not losing their sight and are much more likely to adhere to follow-up recommendations. Orbis continuously collects feedback from users to adapt it to local needs, which vary from region to region.

NGOs have the opportunity to bridge the data divide by adopting intentional practices in data collection, communication, and creation. By gathering on-the-ground data from communities that NGOs serve, effectively communicating findings to drive impact and cross-sector outcomes, and developing data-driven support solutions to address specific problems, NGOs can unlock the power of data to drive positive change. These efforts can contribute to a more inclusive and equitable future, where data is harnessed to benefit all.

**End**