How to Advance an Organizations Data Culture
Implementing analytics at Kaiser Permanente has become more than a function of installing software on computers and telling people to use it. Getting value out of analytics has become a matter of changing the organization’s culture.
That’s the perception of Jeff deRhodes, a business analyst at the organization who has led the charge in getting hundreds within the organization to take part in analyzing varying aspects of operations at Kaiser facilities.
Speaking at the recent Tableau Conference in Las Vegas, deRhodes described what he called an evolution in data culture at Kaiser, and it’s one that he believes most organizations must undergo.
“At Kaiser Permanente, we thought we were going to use Tableau to create dashboards and help drive fact-based decision making,” he said. “What we didn’t expect was the impact that would have on our data culture. Using analytics is altering how we think, behave and work.”
That change in the overall perception of who is responsible for analytics comes from “data democratization,” deRhodes said. “We have more eyeballs looking at the data, and with more people looking at it, we’re getting more insights than we ever did before.”
To change data culture and influence other changes that need to occur to get more insights out of data requires six essential steps, deRhodes said:
Data Strategy. Organizations need to develop a rock-solid data strategy on which to rest their analytics plan. Problems arise when there is no one “source of truth.” Organizations are plagued by “duplications of data and needless duplication of efforts,” deRhodes said. When there is only one source of data, such as that coming out of a data warehouse that contains all significant information, organizations can “get rid of the distractions and focus on insights. That’s not possible when an analyst has to focus on what is the source of truth.”
Data Literacy. For data democratization to truly occur, users need to become knowledgeable about the data they are working with. “It’s good if the people you are working with have an idea of what data types are and what data normalization is,” deRhodes said. Organizations need to increase this literacy by ensuring terms are easy to understand and answering all questions as they arise.
Data Quality. In part, this means weaning people off of spreadsheets and rudimentary databases and having it come through an analytical application that draws from one source of truth. Dashboards, as snapshots of information, are starting points, but questions can arise about the information they’re based on, deRhodes said. Quality of data is an important issue to multiple stakeholders in healthcare organizations, so it’s a key topic to address definitively.
Excel Dependency. Spreadsheets have played a key role in data analysis for years, and part of the challenge in evolving a data culture involves getting users to move to a more effective, more versatile tool, deRhodes said. For example, it’s not unusual for someone with “Excel dependency” to look at a complex analysis of data and ask, “Can you download this to Excel for me?” deRhodes said. “We have to show them the benefits of using different approaches. People are spending a lot of time doing things manually in Excel; we need them to use different tools and use that time to gain insights.”
Rush to Visualization. The graphic display of analytical results can grab a lot of attention, but it may not deliver the desired results. The lack of interest in one of deRhodes’ analytic efforts caused him to rethink the importance of visual display and focus instead on some basic questions to ask users: “Does this help you in your work? Does this change your current workflow in some positive way? Does this cut down the time it takes you to do tasks? Does this reduce your need to manually manipulate data? Are there insights about data that this analysis does not provide?”
Third-party Tools. Some additional applications, layered on a base analytics system, may provide valuable insights. Conversely, analytics staff need to consider whether third-party applications may not add enough value and may only complicate the discovery of insights. Careful analysis is necessary to see if additional tools are helping or hurting.
As data democratization continues, analytics will advance as users become more proficient and their use of data visualization improves. Successes Kaiser Permanente experienced with resource planning dashboards are an example of how the data culture has progressed, deRhodes said.
“Why were they so highly successful? We cleaned up the data, modified processes to make sure we were generating the data we needed. We increased everyone’s data literacy and it was very important that we were getting good business requirements for deliverables. The effort saved us the time of two full-time equivalents in analyzing data. By not continuing to use our old (manual) analytical approaches, we freed up people to really manage their resources better.”