Democratized Analytics: Norton Healthcare’s Visualization ‘Cool Tool’
Norton Healthcare, the regional healthcare provider based in Louisville, found itself grappling with a situation confronted by many data-intensive companies: ungoverned, siloed, inconsistently formatted data that remains stubbornly resistant to analytics and, thus, to providing much value.
Michelle Kannapel is director of business intelligence and financial analytics at Norton, which has 240 locations (five hospitals) and 14,000 employees in Kentucky and Indiana. Looked at digitally, the company is a data volcano, generating a profusion of patient outcome, billing and operational data every day. Kannapel has helped spearhead a five-year digital transformation effort to capture, manage, rationalize and pull insight from the company’s data to improve performance and clinical effectiveness.
This isn’t just a story of how Norton put in place analytics technology. It’s also about how a company learned from a false start and tried a different management strategy for embedding data governance across the organization, one that would create order from data chaos in support of a unified business intelligence capability used by the leadership layer (about 900 end users) of the organization.
Five years ago, Norton made its first attempt to solve the problem of its balkanized data.
“We’ve been through a lot of BI solutions,” she told EnterpriseTech. “We had a chaotic environment, we had different products, we’re a large organization and everybody’s department owned its own data. We used Microsoft analytics solutions, but a lot of our analysts weren’t really being analysts, they spent a lot of time on data manipulation, on report creation. There was redundancy across the system, nothing was standardized, no two reports were the same. There were a lot of misinterpretations based on different reports.”
Norton tasked IT with bringing order to this chaos. But the effort didn’t take root – primarily due to lack of management buy-in but also because data remained within the purview of departments.
By 2015, Norton decided to start over. First it was decreed that there would a single version of truth – that all information would be stored, rationalized, wrangled and accessed from a single data warehouse. Second, there would be a way to access, view and analyze the data and issue reports that were consistent in look and feel and that drew from the same data source. Ownership of the new strategy would be “hybrid”: IT would implement the strategy but it would be led by “data governance champions,” among Norton’s management team, who would hold Norton leaders accountable for using the new unified BI platform.
This is when Kannapel first saw Analytics Explorer (AE), Change Healthcare’s data visualization and analytics tool based on Tibco Software’s Spotfire. Her initial take was that AE “is a cool tool, it’s cool looking.” The graphs and charts it produces have a visual appeal, she said, that makes you want to use the software.
Kannapel initially intended limited use of AE. “We scoped it very small, we didn’t intend this to be our enterprise business intelligent solution, it was going to be just for our team and take the place of some reporting that we had. But then as we saw the power of the tool it’s grown like wildfire in our organization, it’s a fully adopted enterprise-wide BI solution for us.”
This, she said, is because the tool combines “a cool visual factor” along with intuitive analytics that enables a broader number of users to do analytics tasks and exploration formerly consigned to data scientists.
“The appealing thing for me is that my team is not in IT and I have people who want to develop (dashboards) even though that’s not their background,” Kannapel said. “They don’t have to be that very technical person to build a dashboard. It doesn’t take a lot of code. You can take it to that level and we certainly have, but there are things you can do with little to no training that you can still get a lot out of it.”
Self-service analytics, she said, results in better embrace of Norton’s BI strategy.
Kannapel is a case in point. Without a technical background, she found she was able to develop dashboards on her own. “That was a big selling point, the ease of being able to create things in the product, the cool visual factor, the ability to start really high in the data and then drill down until you find the answer,” she said. “It doesn’t take a lot of coding – you can take it to that level, and we certainly have, but you can get a lot out of it with little to no training.”
Tibco CEO Murray Rode said Norton’s experience is typical. “Visualization is how you get the enterprise engaged,” he told EnterpriseTech, “but then it’s nice to have more analytical capabilities behind it.” The idea, he said, is to democratize analytics by automating as much data science as possible while scaling analytics for the enterprise from massive data sets.
Pursuing an enterprise, rather than a desktop, strategy, that emphasizes accessible (i.e., simplified) data exploration combined with data visualization, Tibco claims to have half of the Fortune 500 for customers.
“More predictive use of data – that’s where Spotfire starts to shine,” Rode said. “You can visualize with a rich exploration capability, the ability to start predictive modeling behaviors using statistical models. We’re folding in more sophisticated types of analyses, various kinds of statistical and machine learning algorithms, and then we enable the ability to scale it and administer it for the enterprise.”
Kannapel said the leadership team currently uses about 100 dashboards to track performance across Norton’s lines of business. “We use it across the system monitoring goals we set for ourselves for improving productivity, for costs reduction initiatives, improving outcomes for patients, reducing readmissions, reducing length of stay, reducing mortality, reducing infections,” she said. “It gives us a financial lens, a quick overall look at profitability, how different physicians compare to one another being able to look at their performance.”
She emphasized that in as complex and unpredictable an enterprise as a healthcare organization, opportunities for improvement abound.
“In healthcare there are tons and tons of structured and unstructured data,” she said. “You take millions of lines of data we’ve captured on a patient stay, and we’re able to translate that into some kind of actionable insight. Where before you might get actionable insight in a couple of weeks, now we can translate that very quickly with the product.”
She said Spotfire/Analytics Explorer has helped Norton see through the fog of unpredictability and project daily and monthly staffing resource assignments, such as better forecasting of staff around flu season, in the emergency room and obstetrics. Staffing is a critically important factor because of the shortage of primary care physicians in Norton’s region.
“The question is how to take the current staff we have and make them as efficient as possible, modify their schedules, examine how many patients don’t show up for the appointments, look at whether one doc spends 15 minutes with a patient while another a half hour with a patient with the same condition.”
On the patient side of the equation, Analytics Explorer helped Norton better understand how families consume healthcare through data collected from the patient portal, which showed that women tend to schedule appointments, look up lab results and interact with physicians on behalf of their husbands and children.
“The majority of heavy users are female because females tend to organize care for their families,” she said, “so we’re more engaged in being that center of care for the family, we look at how we can better market to women, get them really engaged with us through our system.”
Other factors the software helps Norton track include no-show patients (they receive extra reminders or, in chronic cases, doctors are encouraged to double-book) and patients who have complained of a bad experience at a Norton facility (nurses and doctors are made aware of the complaints, with the intent of avoiding repeat dissatisfaction).
While Norton’s Spotfire/Analytics Explorer user base is limited to about 900 for now, Kannapel said the company plans to expand its adoption among Norton physicians, which will roughly double the number of end users.