EI and BHC’s systems mapping approach is characterized by three steps (context, change, and action) described in Fig. 1 and evolved from grounded theory methods.
This version of systems mapping was based on primary qualitative data collected via participatory methods (learning sessions and stakeholder engagement workshops) as well as focus group discussions, key informant interviews, and direct observation. The data were used to build causal loop diagrams (“maps”), define inter-connections between issues, and develop narrative descriptions for each loop, opportunity, and proposed action. Each map, when looked at in total, can then provide a detailed description of how things currently work and how things might change to improve the overall health of the city and its citizens. A valuable facet of this approach is its ability to absorb outside data—BHC collected quantitative and qualitative data in each city outside of the systems mapping, and for each set of results, BHC revisited the maps with the new evidence and was able to adjust the descriptions based on these findings. This allowed for triangulation and verification of recommendations and forced BHC to check the system maps for any unexpected interactions or consequences. In this way, the maps were established as living, learning tools.
Typical application of the systems mapping runs between 9 and 12 months in duration. This process is designed to be flexible and can be adapted to different contexts. Elements such as funding or election cycles, sharing opportunities, or project due dates can all be incorporated to develop the most effective timeline.
EI generally suggests a purposive sampling approach targeting participants necessary for achieving (1) a diverse range of informed perspectives on how the city system currently operates, and (2) uptake and support of the final recommendations. Purposive sampling can create bias if it does not recognize some key groups, so EI suggests including the sectors, demographics, and positions needed to accurately describe the structures, behaviors, and mindsets driving current system outcomes. BHC used existing baseline data and relationships with each city to ensure we engaged the right sectors, while also aiming for representation from each of the following: government, civil society, academia, donors, and the private sector. While the number of participants depended on the city, EI suggests an ideal participation rate of 25–40 for each workshop to maximize involvement in the focus group discussions. Since BHC’s maps were living documents, adding new stakeholders throughout the process was allowed if they brought new, relevant perspectives.
BHC also organized a parallel set of citizen engagement workshops called town halls, with groups identified during baseline assessments as vulnerable populations within each city. Their participation was requested separately from the main workshops to give citizens a forum where they did not feel their privacy or freedom to speak might be threatened by those in positions of power.
Prepositioning: Defining the Domain of Each System
Before the workshops began, EI led the BHC team in each city through a series of discussions to:
Define the desired outcome, or “Guiding Star” and “Vision Statement.” Establishing a shared definition of success is the essential first step toward creating a stronger system. The best definitions strike a good balance between inspiring and concrete. Defining a Guiding Star enabled BHC to create a shared definition of what it means to have a healthy system. The team then translated this to a Vision Statement, which describes the desired goal of the system. In the city context, there is often a slogan that can be adapted for this purpose—in BHC’s case, this was usually the Smart City initiative’s slogan or goal statement. Adapting this existing material demonstrated BHC’s commitment to supporting existing city goals (as opposed to criticizing or competing with them) and helped to garner trust from city officials. BHC used the final Vision Statement to keep participants on track throughout the 3-step process.
Frame the system of interest. For the purposes of this work, a system is defined as a diverse set of parts that interact with each other and their environment in ways that are dynamic and often hard to predict, but can be mapped, understood, and influenced. The system boundaries are not pre-defined (as in, the healthcare system, education system, elections system, etc.). EI coached BHC city teams to frame their system boundaries by asking: what is it that we, given our position and potential, most need to know about the current system so that we can understand what is needed to move it closer to the Guiding Star?
Step #1: Defining Context—Data Collection and Analysis Techniques
BHC convened a participatory workshop in each city to collect the qualitative data used to create the building blocks for the first system map: a Theory of Context. Where BHC had done extensive baseline qualitative assessments prior to the Context Workshop, EI coached the BHC team to build from that data for this first step (described in more detail by our Indore and Makassar papers in this issue) [52, 53]. In Da Nang, Vietnam, where BHC had not collected baseline data, the first step was to ask workshop participants to reflect on their own experiences to list key forces that were either “inhibiting” the system from producing health, or “enabling” it to produce health . In both formats for this workshop, participants where then asked to prioritize these forces.
Following the identification step, the key system forces were organized around themes. Participants were broken into small groups of three to conduct a cause and effect analysis on each theme. Each group was given a simple T-chart on poster paper, with their selected theme written at the top. Figure 2 shows one completed example from the BHC city of Indore.
Participants brainstormed and recorded first the causes, or “upstream” factors, that lead to the existence of that force in the system, and then the effects, or “downstream” factors, that this force creates. This was not about consensus, but instead uncovering the diverse perspectives and experiences across the system. Participants were encouraged to check the depth of their analysis through a Structural, Attitudinal, and Transactional (SAT) Framework, provided as a reference during the workshop.
Next, feedback loops were built from these worksheets. These loops provide a powerful tool to understand how different issues may have non-linear relationships with each other. Participants were asked to identify these feedback patterns and build them into causal loops, facilitated by EI and BHC staff (Fig. 3). Having participants build the loops made the final map more reflective of the real system and more readily understood and adopted by key stakeholders.
After the workshop, EI and BHC city teams finalized the loops to develop the context map. EI and BHC used Kumu, an online systems mapping software, to house the data and loops . This platform (with both free and paid options) creates public access to the maps and allowed the team to be transparent in data tracking, storage, and analysis.
To finalize analysis, the team first categorized the causal feedback loops developed in the workshop as either reinforcing or balancing. Reinforcing loops can either be vicious (negative) or virtuous (positive), but they are always self-perpetuating. Balancing loops either stop or limit the initial element, by stabilizing (where something begins as a negative but is mitigated by a positive force) or stagnating (where something begins as a positive but is disrupted by a negative force). Loops are made with factors (bubbles) connected by arrows to indicate the causal relationship. Beside each factor’s connecting arrow is a value sign indicating whether the level of that factor in the system is high/increasing ( +) or low/decreasing (-) (these signs do not denote good or bad).
Next, the team connected the loops to create the full system map. The team defined which loops, key elements, or concepts were repeated most frequently to create a central loop called the “Deep Structure.” This loop holds together the rest of the map. Around the Deep Structure, additional loops are carefully interwoven to reflect the behaviors, interactions, and conflicts identified in the workshop.
The first comprehensive draft system map, the “Provisional Map,” was shared during the community town halls. In small group discussions, participants were asked what most resonated with them and their own experiences, what was surprising or confusing, and what needed to be further explained. Their feedback was then incorporated into the map and each loop narrative. In this way, the context map was always in some state of collection or analysis. Once all essential perspectives were included, it became a “Working Map” and the team moved into the next phase of work, Leveraging.
Step #2: Finding Leverage—Data Collection and Analysis Techniques
The context map served as a tool to identify the best opportunities for impact. Stakeholders were once again convened in a workshop to review the working context map and complete small group data collection and initial analysis. Participants were asked: what is this system functioning perfectly to produce? This question defines where the system is as of the workshop, and opens inquiry into what must change to move it closer to the desired outcome.
During small group work, participants marked the factors in the working context map with colored “flags” to identify types and location of energy in the system (described in Fig. 4). Because factors are framed at the system level, a factor can reflect different types of energy at different levels of analysis.
Next, looking across the flags on the map, groups identified areas where bright spots and energy were clustered and chose which ones they saw as having the greatest potential to shift important behaviors or relationships in the system. Together, they worked through the following analysis process, which is always grounded in what is observed in the map:
What area do we identify as holding high-leverage potential for change in the system?
So what can we learn, and apply, from what is/has already happened here?
What if we could create this shift? What are the direct and proximal impacts we would expect to see (if–then logic within the map)?
How might we connect and strengthen other parts of the system for the greatest potential? Who are the key actors we must engage?
After the workshop, these data from the small groups were synthesized by EI and BHC to identify patterns. As patterns emerged, they were crafted into 3–4 unique but mutually reinforcing “Leverage Opportunities,” visualized as an overlay to the context map in Kumu. All qualitative data collected during this workshop was synthesized into the leverage narratives, and the flags were stored as a map layer in Kumu.
Step #3: Taking Action—Data Collection and Analysis Techniques
This final step again convened the same stakeholders into a workshop. Armed with the context map and leverage opportunities, stakeholders were reconvened to design a set of “Coherent Actions,” which are multi-sector actions that aim to address all of the leverage opportunities relating to one major urban health theme. Focusing on the intersecting patterns within the maps, EI and BHC asked the participants to use the results of the previous workshops to flesh out a set of, “How might we…” questions, which are meant to define those key patterns in the city system that can mitigate, disrupt, or transform the major inhibitors to reaching the city’s vision statement. The BHC facilitators strategically grouped participants by their expertise areas to discuss multi-sector actions addressing similar “How might we?” questions. EI coached the groups to use a facilitation tool called a prototype canvas that fleshed out details for each action.
BHC started the action planning step right as the COVID-19 pandemic began, which limited opportunities to safely share findings and receive feedback as a follow up on the coherent actions. As such, BHC triangulated the action planning workshop results with the project’s other work to complete a draft set of coherent actions. This process is described in Fig. 5.
This triangulation process resulted in additional coherent actions for each city, particularly around multi-sector management, funding, and coordination for activities. It also deepened the descriptions, recommendations, and implementation details for the existing coherent actions developed during the workshop. Data was not stored in Kumu for this phase, but instead was written into a Healthy City Action Plan report. These documents are further described in the city-specific articles in this issue.