The Need for Data Ecosystem Design
As the social sector has become increasingly digitally mature, new digital public goods for creating social impact have started to appear. Ushahidi, the Infectious Diseases Data Observatory, Code for America’s GetCalFresh, and data.org’s Epiverse collaborative are just a few examples of digital products that use data to serve the impact needs of many organizations. However, these social impact products demand a more complex design process than commercial products do because of the complicated ecosystem of data providers, partnerships, and revenue streams required to bring them to life. Moreover, because social impact is at their core, these tools must be built in a way that centers equity and reducing harms, which requires us to co-design such tools with the communities they will affect at each stage of the development process. Despite the unique design skills needed for these tools, there is no formal role for the people creating and evangelizing them. As a result, digital public goods are often built without the skills and resources they need, making them less effective and harder to maintain.
We believe that a new role is needed for the social sector called the “data ecosystem designer” that is charged with creating the data ecosystem that allows digital public goods to thrive and scale in the social sector. These data ecosystem designers are akin to city planners: Just as a city planner must weave together various transportation, housing, and commercial systems to create a city that meets the needs of a large population, so too must the data ecosystem designer weave together various data flows, stakeholder needs, and political processes to create a digital public good that meets the needs of a large population. Until we better define this set of skills needed and the career paths for this role, organizations will not plan for them, individuals will not train in them, and funders will not recognize the critical importance of funding them. The quality and quantity of digital public goods will suffer as a result. At data.org, we have been studying this role through interviews and workshops with leaders in the field. This paper summarizes some of our findings about how this role is defined, trained, developed, and sustained. We believe that better supporting this role will lead to better-designed digital public goods overall.
The Challenges of Ecosystem Design
There are some key differences between digital public goods and commercial products that exemplify the need for this role:
- Impact is key, but it is often hard to measure: In a commercial context, success is defined, crudely, by profits, which is reasonably easy to measure. Digital public goods, however, have an increased onus to deliver a specific societal impact to be considered successful, but such impact is often only seen over a longer time horizon, a horizon that doesn’t often fit to demands for quick measures of success. This societal impact must also ensure that the data and models used don’t perpetuate systemic biases.
- Revenues don’t scale with use: In a commercial context, revenues almost always increase with an increase in customers. In our context, the users of the digital public goods will not always be the ones paying for the service. Unlike private sector models, monetizing digital public goods cannot include the reselling of user data.
- Customers are not uniform: In a commercial context, customers are often very homogeneous. Digital public goods often require the collaboration of users with very different digital maturities and operating models to succeed.
- The definition of “scale” may vary: In a commercial context, while there are various definitions of scale, success is often measured by “units sold”. With digital public goods and their focus on social impact, scale could equate to an increase in a number of users, but it could also represent a move from a pilot to a single major implementer (e.g., a government), a scale to a new geography, or even scaling a tool from one use case to another.
The uniqueness of digital public goods requires a unique skillset to build them and maintain them.
Responsibilities and Skills of the Data Ecosystem Designer
Data ecosystem designers that we worked with cited two main prongs of responsibilities in their role:
Product Ownership and Evangelism: Data ecosystem designers must successfully convey the value of their product and how using it will meaningfully impact the world. They must also recognize what classes of problems the product can address and identify customers with those problems, then convey to partners how to use the tool and what ROI they will get from using it. In interviews, one participant mused that “a product must be usable before it can be useful”, implying the data ecosystem designer needs to understand the needs of all their different users.
Partnership Development: Data ecosystem designers must identify and foster the relationships with external parties that are needed to fill out the data ecosystem as described above. In our interviews, people described having to build technical partnerships, data sharing and governance partnerships, stakeholder partnerships, and funding partnerships – and in each case the partner has to derive a clear benefit from participation. As one participant said, “This person has to be able to bring people on board to their vision, and they have to be able to build trust.” This all requires high EQ, political savvy, great communication skills, and an ability to innovate.
To fulfill these responsibilities, data ecosystem designers need the following skills:
- Data Governance Experience: Data ecosystem designers are building digital products that have data at their core. As a result, they must have a deep understanding of how that data will be used – which data, by whom, at what times, under what conditions. One interviewee also emphasized that data governance also implied an ethical lens. “Because I work with data about human subjects, it’s key to be able to manage the use, security, and privacy of that data ethically.”
- Technology Experience: In our interviews with data ecosystem designers, they all listed technical know-how as requisite, though they felt the data ecosystem designer simply needed to be well-versed in how the technology works. One interviewee pointed out “I think [the data ecosystem designer] has to have the skill to work with technologists, but they don’t necessarily have to have all the tech [skills] themselves.” Designers do not have to come with a STEM Ph.D. or be very hands-on technologically in this role.
- Product Development Experience: Data ecosystem designers will need experience overseeing digital product development, particularly as it pertains to hearing user needs and shaping tools to meet those needs.
- Partnership Mindset: More than any other skill related to technology or product development, we found data ecosystem designers exhibited a mindset for aligning networks of people and creating partnerships that provide value to all parties. It was clear in all instances we studied that digital public goods needed partnerships with other people and institutions to succeed.
- Local Context: On the non-technical front, the data ecosystem designer also must understand the local context within which they are working. They must have some connection to the problems of the people in the environment they’re serving, lest they design ineffective or harmful solutions based solely on their own, non-local context. As one workshop participant point out, it is unlikely that “an app you build for 20 people in San Francisco will work for 5000 people in Kenya”
- Appetite for Risk: In our interviews with data ecosystem designers, many cited needing “bravery” or “resolve” to push new ideas forward in this trailblazing role. Until the role is more standardized, this resilient attitude will play an important role in determining the data ecosystem designer’s success.
- Ability to Code Switch: Data ecosystem designers need to have a range of different product and technical conversations that go beyond just a surface-level understanding with multiple audiences. Code-switching between the language and norms of each group is critical.
What is perhaps most striking about this list is how many skills have nothing to do with technology explicitly. Beyond understanding data governance and how their technology works, data ecosystem designers are primarily tasked with advocacy for the work, building partnerships, communicating a vision, and understanding the human needs of the constituents they serve.
The Data Ecosystem Design Team
From the list of skills above, it’s clear that no one person can do all the work needed for Data Ecosystem Design. Just as the lead city planner works with their city planning team to execute their vision, so too the data ecosystem designer needs a team of technologists, partnership managers, and other skills to create a successful digital public good.
In our research, we found three ways in which data ecosystem designers balance their skillsets through their teams and partnerships:
- Balancing M-shaped skills: No one data ecosystem designer will have all of the skills listed above, but they often are quite deep in more than just one of the skills, giving them an “M-shaped” skill distribution. For example, a data ecosystem designer who came from a law and policy background may be very strong in data governance skills, partnership skills, and code switching. While they may be aware of technology and product design, they may want to bolster their skills there. Data Ecosystem Design teams therefore start by balancing the strengths of the lead designer.
- Balancing organizational constraints: Another dimension data ecosystem designers balance in building their teams is based on the organizational resources available to them. An interviewee pointed out that “[how you supplement your team] will depends on the size and flexibility of your organization, what field you’re in, and what your core product is.” If a data ecosystem designer is working in a tech company to build a digital public good for the social sector, it is a safe assumption that they have technologists and product designers in-house to work with. They may have to build bridges to partnership managers and social sector experts to succeed though. Inversely, a data ecosystem designer that hails from a nonprofit that is trying to scale one of their successful digital products may find the opposite situation. They may have strong social sector relationships and partnerships but need to supplement their technology team to bring the tool to scale.
- Balancing perspectives: Digital public goods must be built for social impact, which means they must speak to all the needs of direct and indirect stakeholders. They must elevate the people they serve in the ways they need without inadvertently inhibiting them in other ways. We found that successful data ecosystem designers made sure to build strong local context and perspectives on their teams. Either through their own experiences, hiring folks with local context, or bringing in people from the field with local context to advise, these designers put the voice and experience of their constituents center stage.
A Path Forward for Data Ecosystem Designers
We feel quite strongly that formalizing this role is important to move toward better digital public goods. Doing so allows practitioners to have a community to share best practices and lessons learned, and to advance the pace of innovation and efficiency in developing digital public goods. We also feel that well-trained data ecosystem designers play a critical role in ensuring that all digital public goods are built responsibly and with the highest standards of data and AI ethics in mind. Indeed, as a new wave of AI funding enters the social sector, we’ll need people trained in these skills to be planning for the long-term sustainability of the products that will be funded, advocating for representative data to be used, understandable models to be built, and products to be overseen so they meet the needs of all stakeholders.
Related Article
The lack of this role in the social sector currently leads to too much fragmentation and the curse of pilotitis: over-focus on short-lived demonstration projects and pilots that have no sustainability because there was no role dedicated to think ahead to what the ecosystem requires for the product to be useful and used in the long term, beyond the initial project that birthed the pilot.
In data.org we have learned to recruit for this role before we even knew what to call it. For example, data.org deployed dedicated staff to support and manage the ecosystem of epidemic and pandemic modeling in our Epiverse collaborative. This program replaces the perils of a fragmented community, where top teams routinely compete with each other for limited funding and produce duplicative ill-supported tools, with an opportunity for teams that used to be rivals in the space to come together around agreed standards and community guidelines to create free-to-use open source epidemiological tools for shared use by the whole community. As a result, the community unites around the best existing software, regardless of who created it, resurrects tools that had been orphaned, and creates new software only where there is a true need, all the while sharing best practices. Our data ecosystem designers’ roles are to bring the different parties together and broker new partnerships, at the same time as also building a network of funders, tech partners and government and intergovernmental agencies to support this growing global community effort. Without the data ecosystem designer role, we would not have been able to pull it off.
Learning from this example and from others, we have three recommendations for the social sector on engendering this role:
- Organizations should acknowledge this role: Almost all folks who have played the role of data ecosystem designer have not had that title. They have been in a role called “Programs Director” or “CTO”, but then asked to carry out these additional, unclear functions. One interviewee in our research exclaimed “When I read this description, I finally thought ‘oh, THIS is what my role is called!’” By raising awareness of this role and the skills needed to carry it out successfully, organizations building digital public goods can begin planning for it and hiring for it.
- Funders should fund this role: Many philanthropic funders have recognized the need for data scientists and AI researchers in nonprofit digital technology creation and have funded organizations to hire them. A natural next step would be to fund the role of the data ecosystem designer. This funding could come in the form of direct funding for a person to serve in the role or through supporting fellowship programs, like Schmidt Futures’ Technologists for Global Transformation program, that place people with these skills in nonprofits.
- Programs should train this role: Just as data science programs arose over the last decade to round out the skillsets of computer scientists and statisticians holding the title, universities should launch programs to train data ecosystem designers. These programs could live at the intersection of technology, policy, and product design programs.
With these three shifts, we can start building capacity in the social sector to bring digital products from the pilot stage through to the level of scale we need for digital public goods. Data.org is committed to championing this role and, in addition to recruiting data ecosystem designers for our own programs, we are developing a charterhood process for other such practitioners in the social sector.
Conclusions
The social sector has made great strides in expanding its data and technology capacity in the last 10 years. To continue that trend to the point that the social sector is solving meaningful problems with digital public goods, we need to go beyond just training more technologists and also define the roles needed at higher levels of strategy. The data ecosystem designer is one of these roles that can bring the sector from pilots to products. The sooner we start prioritizing this role, and developing the career pathways that feed into it, the sooner we’ll be able to see data and AI conceived, developed, and deployed for the greater good.