This post was written by Kyle Alden, geographic information system (GIS) specialist, Bureau for Resilience, Environment and Food Security (REFS), Analysis and Learning Division, USAID.

Over the past year, USAID staff deployed a web-based decision support tool to assist our in-country partners in the 20 Feed the Future target countries in selecting subnational Feed the Future zones of influence (ZOIs). These ZOIs, as articulated by the U.S. government’s Global Food Security Strategy that guides the whole-of-government Feed the Future initiative, are areas where the initiative “will focus and concentrate our resources in discrete geographies and measure progress over time at the population level.” The selection criteria for ZOIs, also derived from the Global Food Security Strategy, focuses on targeting areas with high levels of need (defined as high levels of poverty, hunger and malnutrition), areas with opportunities for agricultural-led growth and areas where there are significant existing or planned U.S. government, donor or host government commitments. Each country team was also asked to select the size of their ZOI based on their estimated Feed the Future budgets and cost of achieving sustainable improvements in poverty, hunger and malnutrition throughout the ZOI.

In order to support our USAID colleagues in ZOI selection, we built a tableau-based web tool that uses quantitative subnational data and a standard method to first suggest some locations that would be suitable for a ZOI and an appropriate size for a ZOI. The tool then allows our in-country counterparts to build and evaluate ZOI scenarios using the same data inputs. We provided country teams with an overview brief that outlined the ZOI selection criteria and the tool functionality, and then provided access to the tool and worked with the country teams to build country-specific ZOI scenarios. While the tool is tailored to the specific task of selecting Feed the Future ZOIs, we have identified four lessons learned from our experience that we believe may be broadly applicable:

1. Clear documentation of the criteria and method was essential for scoping the problem. Referencing one guidance document throughout the process helped to both build the tool and answer any questions as we were working across a wide group of stakeholders. For example, we found that many people — including ourselves — have a tendency to want to consider all possible data sources, but having written criteria helped to scope the decision space since the criteria asked specifically for one indicator of poverty, one indicator of stunting and one indicator of hunger.
2. While we benefited from the criteria and method being standardized, customization of the input datasets was important both for achieving higher quality results, and for encouraging engagement and buy-in from the team in each country. Our standard input datasets were identified because of their quality and availability across most Feed the Future target countries. However, more recent or more granular datasets were often available for specific countries or specific indicators, and our ability to easily substitute those custom data into the model improved the results. This approach to customization became especially important for more subjective criteria, including identifying areas with strong opportunities for agricultural-led growth, which could take different forms depending on the local context.
3. Scenario building and scenario evaluation was key to incorporating qualitative elements into the decision-making process. While quantitative data can help ensure that we were bringing high-quality evidence to our decision-making, qualitative data and on-the-ground knowledge were also important. Our tool allowed users to select areas as a proposed ZOI and then run an analysis to determine whether the ZOI remained well-sized, while also meeting the selection criteria and allowing users to visualize and evaluate tradeoffs. Users could add an area with a strong qualitative justification and remove areas and check the output to ensure that the population size and level of need remained well-aligned with the stated criteria. This iterative process encouraged consultative decision-making and helped users to clearly justify their ZOI selections.
4. The first output can often bias the conversation. Despite providing countless caveats noting that the first output was an example of where a ZOI could be, the selection process often went off course if the first output did not align with the user’s expectations. We suggest finding ways to clearly demonstrate uncertainty in an initial result visually, verbally and in writing.

These lessons, among others, improved our ability to make evidence-based decisions about where ZOIs will be located within Feed the Future target countries. The tool was generally well-received and used to help select most new Feed the Future ZOIs, streamlining our in-country partners’ decision and documentation processes. While the selection process is currently still being finalized, the learning and relationships developed through this process allowed us to take the framework used for this tool and apply it to another geographic targeting problem set. Moving forward, USAID hopes to build on this work moving forward to encourage and simplify evidence-based spatial decision-making.