Predictive Analytics Builds a More Agile Business
As organizations find themselves awash in growing volumes of data, there's a need to focus on actionable intelligence through predictive analytics.
A fundamental goal of business has always been to understand things better and put information to work more strategically and efficiently. However, in an age of swift change and ongoing digital disruption, the stakes are even greater. "For the last couple of decades, businesses have had the ability to run reports and look at past events," said Scott Schlesinger, principal at E&Y's America's Leader for Information Management practice. "But predictive analytics has emerged from the need to look forward and understand how events and business decisions impact the future."
To be sure, organizations across a wide swath of industries are turning to predictive analytics to build a more agile business and tune into key issues, challenges and concerns that might have previously flown below the radar. According to Gartner, 70 percent of high performing businesses will use real-time predictive analytics by 2016. What's more, organizations that use it effectively will increase their profitably by 20 percent by 2017.
Steffin Harris, North American head of big data and analytics for consulting firm Capgemini, said, "Predictive analytics is part of an enormous transformation that is leading to far more granular insights about the market, consumers and trends. It is emerging as a cornerstone for innovation."
How can CIOs maximize and coordinate efforts to use predictive analytics effectively–and reshape their organizations? What technology, skills and culture are necessary to put an initiative into motion and realize bottom line results? And how does the Internet of things enter into the equation? There are no simple answers. As Schlesinger put it: "We are entering a new era that introduces opportunities to gain broader and deeper insights than ever before. Organizations can use predictive analytics to increase operational efficiency, become more competitive, improve profitability and run an organization better from end to end."
Making Predictions Count
Although many organizations and CIOs are eager to embrace predictive analytics, the road to success is paved with more than a few obstacles. It's critical to understand what these tools can achieve for the business but also what's required to produce tangible results. The path may meander into a number of areas: new and improved infrastructure and systems, such as Hadoop or SAP HANA; clouds that can manage and move data more quickly and flexibly; and measurement and data tools, including sensors, beacons and social media streams. No less important: a need to blow up silos and better coordinate efforts across departmental and divisional boundaries.
One organization embracing the concept is Highmark Health. The organization focuses on improving health care outcomes for its family of companies, including the Allegheny Health Network and HM Health Solutions, which provide health care for 40 million Americans in 50 states. The health network has more than 5.3 million plan members and it is the third largest integrated delivery and financing system in the U.S. One huge challenge is balancing the sometimes competing needs to improve outcomes and quality of care while managing costs. "The vast majority of the clinical information we encounter is in the form of unstructured data or textual data. It contains a lot of predictive power involving the business and clinical problems we’re attempting to solve," said Mark Pitts, vice president of enterprise informatics, data and analytics.
Highmark Health has turned to SAS Enterprise Miner and Text Miner to map and analyze likely outcomes a patient faces based on factors such as symptoms, health history and demographics. This means, for example, that as the health care provider looks to reduce hospital re-admissions, it is able to examine wide ranging data fields–such as whether a person has diabetes and his or her age and weight–and predict the likelihood that the patient will require additional care. "We would intercept a discharge note as it is produced in the system and apply a predictive score. The benefit is that we are able to focus resources on the individuals most likely to be re-admitted," Pitts explained.
The company expects impressive results from the initiative. It hopes to reduce patient re-admissions, improve both the speed and quality of care for patients, and save countless hours of practitioners' time previously spent on reviewing charts and records. This, in turn, will help the enterprise identify millions of dollars a year in previously undetected medical risks. What's more, Highmark is looking to expand the data initiative by identifying patterns that lead to particular health problems or diseases and adjust care and treatments accordingly. "We are able to build increasingly powerful predictive models that redefine health care and the fundamental way we approach the business," Pitts said. "We are able to balance quality and costs more effectively."
By the Numbers
Predictive analytics has growing value in industries as diverse as retail, hospitality, financial services, manufacturing and even within organizations for human resources and retention purposes, Harris said. Maximizing results requires organizations–and CIOs–to focus on building an infrastructure that supports real-time streaming data, whether it originates from POS terminals, beacons, machine or smartphone sensors or social media. There's also a need to add data science skills and gain deeper insights into emerging technologies–all while understanding overarching industry challenges and marketplace trends. All of this, he said, falls somewhere between art and science. "You have to bring together diverse groups within the organization–and better understand interactions with partners and customers–in order to build truly advanced predictive models."
Best practice organizations, he added, typically develop an advanced analytics competency center within the organization. Ideally, it ties together four critical areas: IT and data infrastructure; BI and analytics tools; business support (including developing the necessary data science and other skill sets); and establishing a powerful and comprehensive governance model. Schlesinger said it's vital to recognize that "information for information's sake" is not the end goal. As organizations find themselves awash in massive and growing volumes of data, there's a need to focus on actionable intelligence. "The amazing insights come when you have a clear understanding of your opportunities and goals and what benefit predictive analytics delivers."