Strong data strategies can enable business decision-making — but what happens when a data strategy overlooks necessary collection and analysis?
As data and analytics become a core business function, building a data strategy adaptable to disruption requires planning for innovations with a focus on the end user. At the Department of Veterans Affairs, the agency had to revamp its data collection efforts after noticing its shortcomings in collecting data on gender and orientation.
In part because of the stigma around not identifying as cisgender and heterosexual, "we have not historically done a good job in collecting data on sexual orientation and gender identity," Michael Kauth, director of the Lesbian, Gay, Bisexual, and Transgender Health Program in the Office of Population Health, Patient Care Services at the Veterans Health Administration, a sub-agency within the VA, said during a panel Wednesday.
The information wouldn't show up on a veteran's records unless they were being treated for their gender identity or sexual orientation as a disorder. And, in some cases, veterans decided not to provide that information out of fear of discrimination. "There were lots of reasons to keep that a secret from healthcare providers again because it leads to being punished, it leads to losing civil rights," Kauth said.
Ultimately, the VA had to correct the stigma around providing information about gender identity and sexual orientation, and deal with a backlog of missing healthcare information professionals needed to provide the best care possible.
To start gathering the information, the department now aims to include sexual orientation and gender identity data fields as requirements on its electronic health records system, according to Kauth. The Government Accountability Office recommended the change in a report and the VA set up a working group to create those fields.
No matter the context, data strategies at most organizations will have to adapt to changing circumstances or the need for additional data.
When thinking about data strategy, Alan Henson, principal at Pariveda Solutions, recommends starting with considerations for the end users, such as VA did with the new data fields. Once the team is familiar with the consumers of the data project, it can design a strategy around serving those needs to avoid ending up "with a bunch of data that doesn't serve a real clear and concise purpose."
For organizations that may have deployed a data strategy, but now realize gaps or flaws, the first step toward course correction is to "take the time to stop and look," Henson said. Assess the gaps based on the outcome desired with the business use cases in mind.
"From there you need to get a plan of attack," Henson said. "Oftentimes, companies have all the right people in place to solve some of these problems, they just need to get them together, and then start to execute on that plan and make it incremental."
Building a better data strategy
While collecting the data is a good starting point, the VA also recognizes that its methods will have to evolve and change to stay relevant to shifting identities and terminology.
"There's a lot of interconnected systems, there's a lot of work on backends that goes on," Danielle Thierry, content strategist at the U.S. Digital Service for the VA, said on the panel. "We're going to have to keep evolving in an agile way and also as we see people's identities are changing and the terms that we use and the ways that we express that [are] changing and broadening."
But requiring veterans to fill in the additional fields will begin to prompt more accurate data collection and, ultimately, healthcare tailored to each veteran's needs.
"Talking to statisticians and researchers, they really drove home the point that those different fields also helped with the accuracy of data collection as well," Thierry said.
Nearly half of companies want to improve data quality and analytics over the next two years, according to a MIT Technology Review Insights report released in April. But it starts with a strategy ready to adapt to the improvements.
An incremental plan recognizes that the strategy will change as the future unfolds, and is flexible based on business needs. Henson recommends making the plan less detailed and more high level as it looks further out into the future.
"In the software development world they call it agile," Henson said. "Having a higher-level plan helps keep that roadmap focused in the same direction. The way you keep it flexible is you're periodically checking it and pivoting as necessary."
Further development of AI and machine learning, the increase in data volume to compute at the edge, and changes in data privacy are all disruptions businesses should be preparing for, according to Henson.
"It's really important that the data strategy has support at the highest levels of the organization in order to be successful," Henson said. "Those organizations that leverage blended teams between data scientists, technologists, [and] business users typically have the most success because you're getting all the players in the room to make decisions on what that strategy should ultimately look like and how best to meet the business's needs."