Plan for data analytics success with these tips from Geisinger

BOSTON—Analytics is not a strategy, said Nicholas Marko, MD, Geisinger Health System’s chief data officer, speaking at the Big Data & Healthcare Analytics Forum.

It seems self-evident, he said, that the people managing data analytics projects aren’t necessarily the ones who should be responsible for defining and prioritizing strategy.

“Data strategy is one piece of a bigger thing,” Marko said. “There is a larger ecosystem to think about when using information.” That’s why Geisinger created an enterprise data strategy with representation from across the organization.

“Strategy is a dynamic and evolving process,” Marko added. People tend to get together to create a roadmap “but looking for that static guidance, governance document is counterproductive. It has to be a living thing.”

A roadmap typically looks completely different six months after its creation which is natural, he said. There’s no one right way to establish a data strategy because each organization has to do it in a way that makes sense for them. Marko also noted that there is not finite end point to the process.

Failure in analytics doesn’t happen in the predictive modeling part, but at the beginning or the end. Organizations find out they have answered questions that weren’t necessarily the questions they needed answered. “We try to focus on making sure our questions are clearly defined and when we produce something, we have someone to hand it to who can do something with it,” Marko said.

Reporting and dashboarding is different than analytics, Marko explained. “Dashboards just summarize data. Analytics is the part where you’re taking data, putting it together in a novel way and using it to generate some value.”  

Marko broke down the planning to three segments: vision, administrative and technical.

Your analytic vision answers your key questions. “You can have a great team but without clear questions you will be disappointed with what they produce.”

Organizations also should determine their value proposition or why they’re doing this. To save money, to improve their reputation, to improve care? “It’s probably some sweet spot combination,” Marko said. But, be very clear because this is expensive. “There is lots of cool stuff you can do with data, but we all have limited resources.”

On the administrative side, you need a commitment from your leadership. “If your senior leaders don’t think you need a data environment, you’re not going to move on this.” Marko said you don’t need much to do effective analytics but you do need to get your priorities synched up. Also think about who is managing the vision. “It’s good to have someone in charge. Rule by committee is great but not particularly agile. You need to drive forward otherwise you have a bunch of people with good ideas that guarantee they never get anything done.”

Successful data analytics requires a good technical lead, Marko said, recommending that organizations take an honest assessment of their in-house expertise. Not all the analysts you like have the expertise to do what you want done, he noted. You should also evaluate your infrastructural capability. Geisinger made some upgrades to support analytics and provide more flexibility on the back end.

Underpromise and overdeliver, Marko advised. “Don’t tell anyone all the cool stuff you think you can do.” Use caution with vendors and consultants, he said. “They can be very helpful when used appropriately but there are lots of people who will sell you expensive stuff that doesn’t make for an analytics enterprise.”

Build for being adaptive because “things will always be changing. You cannot build a static solution. Think about how you do what you need to do today and make sure you can do what you need to do tomorrow. Planning for success in two years is what separates the wheat from the chaff.”