The power of data ecosystems

As the COVID-19 pandemic waned, many people, including in the Bureau of Economic Analysis (BEA) within the US Department of Commerce, were wondering when the economy would rebound. But waiting until companies’ sales reports were delivered seemed like such an antiquated way of analyzing economic recovery. So, the BEA decided to look outside to a third party for data, specifically credit card purchases.¹ They reviewed the data industry by industry, looking at groceries versus luxury goods, and compared current purchases with pre- and mid-pandemic numbers. As a result, they had near real–time data to determine how well certain parts of the country were rebounding economically, down to the zip code.

This is an example of a data ecosystem in action. By thinking, sharing, and using data from outside traditional boundaries, an organization can connect the dots between the data sets and gather a more holistic view of what drives outcomes. Whether academic institutions, consortiums of government agencies, or commercial companies, the partnership options for an organization can be nearly limitless.

How do data ecosystems create value for participants?

Data ecosystems can offer the opportunity to unlock the next level of actionable insights. Generally, these ecosystems allow organizations to achieve richer insight and serve citizens more effectively. For example, organizations with high data maturity are currently utilizing insights to drive operations, strategic decision-making, and oversight.² That said, many organizations can benefit from considering ways to expand their network and what external information could be beneficial.

Data ecosystems promote openness and transparency among participating entities through a shared set of values. When a group of participants share a common goal, say improving public health outcomes or reducing commute times in a city, participants can win when they share data. Even individual citizens have shown willingness to engage with data to help improve their communities.³ As government agencies think about creating these data ecosystems, a benefit of sharing their own data is not only to enlist the help of a broader community on mission challenges, but also to help demonstrate good faith so that other participants will continue sharing their data.

Technology advancements in the cloud make it easy to create these data ecosystems. Data can be seamlessly shared with the proper security and governance in place, reducing the manual work that historically was required to consolidate and share data.

When does a data ecosystem make sense?

More data and lower costs are likely only worthwhile if they help create some mission value. The cost and effort to establish and maintain a data ecosystem may neither be sustainable nor economical if participants do not capture enough value. Data ecosystems can create mission value in two ways:

• You need more data than you own. Complex problems like public health or local economic vitality can be difficult to capture in any single data set. Rather, they often need multiple looks from multiple perspectives. For organizations grappling with such problems, it can be difficult to own all of the data needed to understand the problem. A data ecosystem can help bring all of the relevant data to bear regardless of who owns it. This is exactly what researchers at the National Institutes of Health did with the Helping to End Addiction Long-term data ecosystem, bringing together data from academic researchers, health care providers, and communities to help combat the opioid crisis.⁴
• You need more talent than you own. In other cases, government may have a monopoly on the data but lacks some of the technical skills needed to make sense of it. In these cases, a data ecosystem can help not by bringing in new data, but by bringing new talent with new skills to bear. This is the problem that faced the Federal Railroad Administration (FRA). The FRA needed a faster way to inspect the tracks used by high-speed trains. Because of the speeds involved, these tracks need frequent and thorough inspections, which can be time-consuming and costly. The FRA had all the data on tracks that it needed, but it lacked some of the image processing and machine learning expertise needed to make sense of it. By turning to a data ecosystem of industry and academia, FRA was able to devise a new method of track inspection that could both save time and improve safety.⁵

Build the technology, build trust

Regardless of the reason driving it, sustainable data ecosystems can require participants to navigate competing motives. While competition between agencies may occasionally be the source of incentives against data-sharing, more often in the public sector, the motives are balancing the risk and reward of sharing. Public organizations want to accomplish its mission and thus have an incentive to share. But in many cases that can be overshadowed by incentives to protect sensitive data from even inadvertent compromise. Breaking the tug of war between these incentives can take trust between all players that everyone will protect data as if it is their own.

To sustain a data ecosystem, then, may not only take interoperable technology, but also high levels of trust across the participating entities.

Technology to ease data-sharing. Data privacy and security are foremost considerations for any organization looking to establish or participate in a data ecosystem. The 2018 Federal Data Strategy outlined principles and practices that provide a governmentwide vision for how agencies should manage and use government data.⁶ Alongside this strategy, regulations around data security and privacy are continuously being established to set guidelines for exchanging data in trusted environments. With a collaborative focus on data usage, federal and local government agencies are increasingly investing in technology and frameworks that intrinsically make sharing more achievable than ever before.⁷ The development of cloud, automation, and artificial intelligence (AI) technology provide the embedded framework to help better break down data silos and centralize the management and governance of data, while providing a more seamless way to share information. In this sense, the current environment of policy and technical innovation could be optimal for chief data officers (CDOs) to accelerate their engagement in data ecosystems and allows them to respond to the rising need to leverage data for faster and better decision-making.

Steps to improve trust. Adopting new technologies and data standards can help accelerate sharing, but trust will likely need to be continually maintained every day. For example, in recent years, there has been greater focus on ethical and trustworthy use of data in decision-making and operations.⁸ One key result of this focus on trustworthy data was finding that data ethics need to be continuously reviewed, especially as AI capabilities continue to gain traction, rather than simply being reviewed once. As such, many government entities are already preparing for this scrutiny, including the US Department of Defense, whose 2020 Data Strategy explicitly includes data ethics as one of its eight guiding principles.⁹

While trust among participating organizations is an important requirement, there will likely still need to be a sufficient governance model and policies in place for the data ecosystem. And although technology is rapidly changing to allow for more sovereign data connections, participating organizations should understand and adopt governance rules to safeguard interoperability, data privacy, confidentiality, and security. These rules should be agreed to by all ecosystem participants with sufficient visibility and enforceability by the other members.

An organization’s culture can determine how willing it is to adopt new capabilities and adapt to changes in its operating ecosystem. With expanded data access could come new insights that may challenge assumptions, existing processes, and operating norms. Likewise, with the ability to accelerate the decision-making process, an organization may be required to review how they view and manage risk.

Steps towards change

The power of a data ecosystem to bring new data and new skills to bear on problems can make it a great choice for government organizations who often have to tackle some of the public’s most intractable problems. While data ecosystems can be complicated, a few considerations can help organizations get started:

1. Understand the players and set a common goal. The foundation of a sustainable data ecosystem starts with an understanding of desired outcomes for participants. If different participants are pulling toward different goals, the data ecosystem could struggle to hang together. Mapping current and proposed stakeholders, their motives, and the data they can bring to bear is an important first step in establishing a thriving data ecosystem.
2. Assess the gaps. With a common goal selected, CDOs can inventory their organization’s data sets and skills. They can then assess where gaps exist that require additional data, additional skills, or both. These questions could be instrumental in selecting both the right participants and the right technology for the data ecosystem.
3. Find the right data platform. The security, privacy, ethical, and technical requirements of the future state should be considered as well as a determination on how to identify the right data platform to meet desired requirements. With this baseline understanding in place, CDOs can begin to specify decisions needed and investments required to establish the desired data ecosystem.
4. Design a minimum viable product. CDOs can engage in small-scale ecosystems by aligning on what data can be shared (sourced or supplied), who the trusted ecosystem partners will be, which collaboration model is optimal, and experimenting with the technology available.

Data ecosystems have the potential to drive new insights for the benefit of not only government agencies but also the citizens they serve.