When Swedish Medical Center, a Seattle-based network of five hospitals and outpatient facilities, began implementing a consolidated data warehouse and business intelligence report tools in July 2012, project leaders quickly found that the technical aspects of the implementation would be the easy part. Much harder would be the task of building support for the reports among the medical system's nearly 1,000 physicians.
The challenges of getting non-technical users to adopt BI systems and make them a part of their
You can't move and do analytics unless you have some sort of organized platform first, but we weren't naïve enough to think this was the end goal.
David Delafield, CFO at Swedish Medical Center
At the TDWI BI Executive Summit in San Diego, David Delafield, CFO at Swedish, explained how he helped his organization jump these hurdles. He said top executives decided a little more than a year ago that the system's various hospitals and medical offices lacked a common set of performance metrics, which made it difficult to see which providers were performing well and which needed improvement.
The organization began by building a large data warehouse that collected information from 20 offices throughout the network and put in place analytics layers on top from SAP and Microsoft. But Delafield said he knew this was not the end of the project.
"You can't move and do analytics unless you have some sort of organized platform first, but we weren't naïve enough to think this was the end goal," he said, adding that only a small fraction of the intended clinical users had the technical skill or interest required to use the system.
The next step was to put in place a user dashboard from Tableau. Delafield said this simplified reports and made it easy for doctors to log in and view their performance metrics. However, making things easy for the users wasn't enough. Few doctors viewed the reports available to them because they didn't see how it was relevant to their jobs.
Next, Swedish tried to tie physician pay incentives to their performance on the metrics being tracked by the system. Delafield said this is where the organization really ran into trouble. Physicians who had mainly been ambivalent to the system were now downright hostile. They didn't like the idea of having their pay tied to the IT department's idea of good care, which was how they perceived the system.
The situation was resolved with a simple solution. Doctors were given medical scribes, data entry specialists who electronically document information during patient visits, to enter data for them. This improved documentation and reduced the burden of data entry on the clinicians. It also allowed doctors who were delivering quality to easily show that they were performing up to standards, which most supported. Eventually, doctors started viewing the reports and making efforts to improve their performance.
"The governance process [allowed us] to have real transformation," Delafield said.
One way that many IT departments are trying to manage the workflow and governance changes that go hand-in-hand with BI projects is to use agile development practices. Chris Mills, director of engineering at Cobalt Automotive Market, a division of Automatic Data Processing Inc., said agile practices helped his IT department roll out a major change to the business's data warehouse without causing disruption to users or reports. Agile processes focus on iterative and incremental development. It requires face-to-face interaction between the IT and business departments during development and stresses the value of getting a usable product in the hands of users as quickly as possible.
READ MORE ABOUT ORGANIZING BUSINESS INTELLIGENCE REPORTING
Learn about the value of creating a data-driven culture
See how other businesses are using agile development for BI
Read about how to align IT and business objectives
Cobalt provides market reports to car dealers. The reports detail the types of cars consumers search for online, the market price of specific vehicles and other industry measures. But Mills said anytime a customer wanted a new report before the implementation of agile practices, the process would take months. Mills would receive the request then hand it off to the senior database architect who worked with a developer who uses extract, transform and load tools to build the capability. The whole thing would be tested. Finally, depending on results of the testing, the whole process might have to start over.
In 2010, the company began transitioning to agile development practices to streamline the process. Now there is a fulltime product owner on the analytics team who works with business departments to determine what users need. New reports are tested as they are developed, which saves a lot of time at the end of the process, Mills said. Simultaneous testing of newly developed products is a key tenet of agile development.
The true test of agile came when the IT department wanted to move the company's data warehouse from Oracle to Greenplum. There are differences in how the two store and sort data, which could have altered many of the BI reports Cobalt was producing. But the analytics team was able to quickly test each report and make updates as necessary to ensure that nothing changed. Mills said once the data warehouse switch was completed, none of the report users noticed any difference.
This was first published in September 2013