Duke University Health Systems takes data democracy to next level

Duke University Health Systems (DUHS) has more than 22K employees and is the second largest private employer in North Carolina with a network of health services, hospitals, support services, and biomedical research. The DUHS IT team often works with the Duke University IT group and other internal organizations. Through the dedication of a small group of employees, there is now a growing data analytics community where data democracy, knowledge exchange, and team innovation thrive.

Tableau Desktop and Tableau Server were integral to leveraging the massive amount of information across Duke. There was a Tableau User Group that brought together Tableau users at Duke to share best practices and new ideas, but no group existed to address the underlying data questions that feed Tableau in the first place–what data exists, who can access it, and how can it be better.

Claire Howell, Clinical Data Strategy Manager on the Duke Cancer Institute (DCI) IT Clinical Data Strategy Team, and Rebecca McDaniel, Duke University IT Solutions Architect learned this first-hand when they worked on a data project together. Both had a passion for Tableau but realized they knew different pieces of information about data that others on their teams didn’t. This insight put the spotlight on the unintentional information silos across departments, and an opportunity to centralize information for data analytics employees.

As McDaniel put it, “We believed that before you can fully tap into the abilities of Tableau, you have to first create a foundation of information accessibility. Accessibility is empowerment. By empowering individuals to access data, they can then work together to maximize the use of those tools.”

Joining forces with a larger team and surveying the analytics teams across Duke, Howell and McDaniel determined that a department-agnostic space—a data analytics community—was the strong foundation needed to help break down those barriers that existed in accessing data across the organization. “We tried to find this balance of not wanting to be too structured because we want it to be open source, but for people to feel empowered to make decisions themselves and to run with ideas,” Howell said.

What made the community even more effective for data accessibility was that it included both bottoms-up participation and executive sponsors to champion programs, provide transparency about decisions affecting data analytics teams, and shine a light on leadership and community priorities.

Additionally, the community intentionally partnered with existing groups rather than duplicate efforts. For example, instead of creating a new Tableau User Group within the Duke Analytics Community, Duke Analytics Community members are routed to the Tableau User Group.

With the perseverance put in, the data analytics community has become a success on many levels—not just for its members, but for amplifying data culture across Duke. Data accessibility is woven into the group’s DNA. As Howell put it, “It's been amazing how people have been so generous with sharing information. Leadership is coming in and allowing that space to be the stage to hear from analysts and try to get feedback on how things can be improved or better. It's been really rewarding.”

Some of the major results included:

• Centralized communication for data analytics announcements, upgrades, and other information
• Established intentionality for equal access, including rotating moderators and a crowd-sourced public task board
• Improved project allocation and resources through executive support
• Provided open community discussion forum and channels for code sharing, SQL expertise, and other data analytics topics
• Highlighted shared resources, such as internal data training programs and community member-sourced style guides

Reinforcing the power of the community beyond data analytics, the core team launched a logo contest for its ever-growing members without high hopes for a big response. Instead, ideas were submitted from more than a dozen members. “If we create the space for it, it turns out people do want to share, and helpful members are happy to respond. Seeing that happen organically doesn’t feel like credit to us, it’s the group,” McDaniel said.

This is just the beginning for the Duke Analytics Community. In addition to growing its membership, it’s serving as a best practices model for putting information accessibility into action, at Duke University and beyond.

Acknowledgements: The Duke University Tableau User Group was created by Andrew Stirling and Phillip Overpeck. Howell and McDaniel thanked Stephen Blackwelder, Michael Pencina, Dan Cantrell, Nelda Bradley, Wil Burgon, and Ursula Rogers for their partnership establishing the Duke Analytics Community.