Lessons for Upskilling Earth Observation in Resource-Constrained Settings

Geospatial data is no longer a niche capability. It’s an essential ingredient for climate adaptation, disaster response, and sustainable development.

At Caribou, we’ve worked with partners across the geospatial sector to design initiatives that help local and regional humanitarian actors develop practical geospatial skills that inform and enhance their day-to-day decision-making.

But one question we’ve returned to again and again, especially through our work with the UK Humanitarian Innovation Hub (UK HIH), is:

Caribou has seen the power of digital upskilling in action. Through the Mastercard Strive program, we support small and medium-sized enterprises (SMEs) and startups to strengthen their digital capabilities. Our research on youth livelihoods has also shown how accessible relevant training can improve productivity and expand economic opportunities. The same is true for geospatial data, which is increasingly critical for navigating humanitarian and climate-related challenges. 

We encountered this directly through our work with the UK HIH, co-developing the Geospatial Data and Technical Support for Humanitarians in East Africa initiative, a technical assistance program designed to equip local and regional actors, often operating without in-house expertise, with the skills to apply geospatial data in day-to-day decision-making.

To broaden access and promote inclusion beyond well-resourced INGOs, who often benefit from in-house Earth Observation (EO) teams or global partnerships, we launched an open call for humanitarian organizations to apply for tailored EO skills training. The response was clear: we received 88 applications, underscoring just how much demand there is for practical, hands-on support among smaller, under-resourced organizations.

We knew that delivering meaningful support required more than just demand. It meant selecting partners who were motivated, well-positioned, and ready to engage. Applications were screened for eligibility and then assessed by a panel of seven reviewers. We looked for organizations that showed a strong commitment to using EO, a clear understanding of how it could address operational challenges, and the potential to apply what they learned in ways that could benefit others in the sector.

After five independent reviews per application and a series of follow-up interviews, four organizations were selected:

Each organization received up to 12 days of tailored technical support, combining both virtual and face-to-face training. The curriculum—developed and delivered by Humanitarian OpenStreetMap (HOT), MapAction, and Advanced Geospatial Solutions (AGS)—was customized to each organization’s needs, with a strong focus on using open-access data and tools. Through this process, we identified six key lessons for designing EO upskilling that is inclusive, scalable, and grounded in operational realities.

We found that training worked best when it was tied directly to participants’ day-to-day responsibilities. At UNHCR, for instance, staff applied tools like KoboToolbox to map refugee infrastructure and service delivery, directly embedding learning into operational tasks. CEFORD participants used their training to identify and map agricultural activity, helping visualize land use for program planning. For LM International, along with MapAction, we shaped the training around their ongoing program work, selecting pilot data collection projects aligned with its existing work.

Not everyone learns the same way, and we saw early on that flexibility was key. We used a mix of face-to-face sessions, asynchronous digital content (like short video explainers and downloadable guides), and mobile-friendly channels like WhatsApp groups. This helped participants learn at their own pace and revisit key material when needed. The capacity-building consultants shared core theory up front to provide a shared foundation for participants.

At LM International, this approach meant participants came to the training with a solid foundation, ready to map, collect data, and start analyzing. At KRLON, the hybrid model worked especially well: participants arrived well prepared and jumped straight into building surveys in KoboToolbox and mapping data with QGIS. One local group even adapted the training tools into their own IT classes for young people in the Kyaka II Refugee Settlement. To support this, HOT co-created short video tutorials that the group could keep using and sharing.

We didn’t want training to stop when the sessions ended. We wanted it to spread. That’s why we designed the program with organizational networks in mind, looking for opportunities to support not just individual teams, but the broader communities they were already connected to. 

CEFORD’s strong connections with government partners, other NGOs, and refugee communities meant the training was useful to a wide network. At UNHCR in Uganda, we brought in staff from the Norwegian Refugee Council (NRC) and the Danish Refugee Council (DRC), creating space for inter-agency learning. With KRLON, it was great to see other organizations from around the Kyaka settlement joining in — sometimes even informally — because they’d heard the training was worth attending. That kind of ripple effect wasn’t incidental. We designed for it.

We went in to build geospatial skills, but the moment the capacity-building consultants started working with teams, it became clear that the challenges ran deeper. For some organizations, just storing and managing their data was a hurdle. Others were dealing with inconsistent workflows or struggling with the basics of digital infrastructure.

At CEFORD, those gaps became opportunities. Together with AGS, CEFORD worked to set up cloud-based storage, simplify data collection processes, and build stronger habits around documentation and data security. One moment stuck with us: a government minister visited during the program and asked about CEFORD’s data practices. The team answered confidently, not just because of the training, but because they had taken ownership of the systems behind it.

We knew that the organizations most likely to benefit from the training would be the ones most eager to engage, so we built motivation into the selection process. In our open call, we didn’t just ask about technical needs. We asked applicants to show their understanding of how geospatial data could support decision-making in their context and to explain the time and resources they were prepared to commit. We looked for signals that training wouldn’t stop with one team. Did applicants see how EO could help tackle the challenges they faced? Had they thought practically about how to embed it into their work and share it across their networks? That approach meant we weren’t just building skills. We were laying the groundwork for knowledge to flow through existing partnerships and peer groups.

That approach worked. Organizations threw themselves into the training once it began. CEFORD, for example, showed early on that they were serious, and they followed through, dedicating staff time, asking thoughtful questions, and testing tools in real-world scenarios. In the end, the most sustained impact didn’t come from the flashiest applications. It came from the teams who were ready to invest in the process and make the most of what was on offer.

We didn’t treat training as a stand-alone solution. It was part of a broader ecosystem of interventions co-developed with HIH in response to common barriers to EO adoption in the humanitarian sector. This ecosystem includes:

These tools address structural constraints that one-off training alone cannot solve. Designed for low-tech users, they offered practical, scalable ways to extend learning and support better, more confident decision-making across different entry points.

The Geospatial Data and Technical Support for Humanitarians in East Africa initiative demonstrates that capacity building, when embedded in a broader ecosystem of tools, trust, and shared ownership, can deliver impact far beyond its initial footprint.

That is the future we are working toward at Caribou, and we hope others will join us in building it. Reach out if you’re interested in designing similar support.