In the insurance business, most technology change has been characterized by evolution, not revolution. However, some technology advances have been transformational, and big data is poised to be counted among them.
"Big data is a major part of the digital revolution, simply based on the sheer explosion of data available to companies," says Denise Garth, partner and chief digital officer at SMA.
Big data provides an unprecedented wealth of new information to help insurers achieve longstanding goals. "On one level, insurers are targeting traditional objectives—how can we be profitable, how do we manage risk, and so on—and being more precise in doing so because of the growth of information and increase in analytic capabilities," says Mark Breading, SMA partner.
Additionally, big data is transforming the relationship between customers and businesses—including insurers. "Policyholders' expectations for their relationships with their insurers are being set by competitors outside the industry—the Googles and Amazons and Facebooks that have a higher level of data mastery and are able understand what consumers want and what their needs might be. Insurers are going to need to develop that same level of big data mastery and gain the same level of insight," Garth says.
Big Data, Big Challenges
Leveraging big data starts with coming to an agreement on what it is. Although generically defined as collection of data so large and complex that it becomes difficult to process using traditional approaches, in the P&C context it often is used as a catch phrase to characterize any large amount of information that lives outside a traditional database.
"There are a lot of different sources of third-party data, both public and within the industry, that have come online, and the ability to move large amounts of information has become much easier," says Justin VanOpdorp, senior vice president in analytics at Lockton. "For us, big data really means digging through the wealth of data that's out there, uncovering trends, and identifying where we see opportunity to provide useful and practical information that helps make our business and our client's business better."
"Insurers are sitting on a virtual gold mine of information," says John Anderson, senior managing consultant in IBM's North American Strategy and Analytics Practice. "Claims activity, underwriting activity, sales and distribution activity. All of that information should be collected and compiled, and that's beginning to happen as insurers go through transformational changes to their claim systems, underwriting platforms, and even to their distribution capabilities."
The world of big data is only going to get bigger as the Internet of Things captures more information and moves it from machine to machine. "Of the industries across the entire spectrum, insurance is positioned to be a winner in the big data environment because of data coming from in-vehicle devices, medical devices, and more," says Joe Caserta, president of analytics consultancy Caserta Concepts. "If insurers can find a way to take this data, digest it, and analyze it into actionable results, they can drastically improve their results."
This increasing velocity and expanding volume are two of the "three Vs" of big data that challenge insurers. Technology approaches that provide both storage capacity and processing power to address volume and velocity have come online, such as Hadoop, which has become nearly synonymous with big data analytics.
Caserta believes that using traditional, structured data warehouses to support analytics will become a thing of the past, even in the insurance industry. "We are definitely seeing a migration from traditional relational databases to this big data paradigm," he observes. "The most important reason is cost. Traditional databases are designed to enforce integrity when doing transactions. For analytics, you don't really need that—it's unnecessary overhead."
The third V is variety. In contrast to the structured, transactional data that insurers are accustomed to dealing with, as much as 80 percent of all data available to insurers is unstructured, including images, audio, adjuster notes, and social content.
"We've had unstructured data for years, but we haven't fully figured out how to use it," says SMA's Garth. "A big focus of the analytic vendors has been developing tools to mine this unstructured information, pull the data artifacts out, and combine it with structured data for analytics."
"Unstructured data is a challenge," says Jim Wilson, vice president, strategic analytics at XL Group. Wilson says that the company is currently "kicking the tires" of big data and building on its experience with solutions from SAS. That includes creating a data lake, a large repository designed to hold data in its native state, rather than being optimized for traditional analytics.
"We use SAS as a discovery ground where we take on data feeds from different organizations, mash it together, and conduct iterative analyses," Wilson says. "Rather than develop the traditional stack of ETL, warehousing, and storage, we are exploring how to change that by discovering first, then adding structure when we need to."
Next page: How much data is too much?
Understand the Why
Storage and processing power aside, as the volume of information grows, insurers struggle organizationally with how much data is too much. As the volume of information grows, insurers struggle organizationally with how much data is too much.
"People used to say I want all the information, but the problem was they couldn't handle all the information," Anderson says. "Even now, with transformational changes and new technology, this can be a problem unless a company is willing to work on the data governance component."
As the saying goes, the way to eat an elephant is one bite at a time. SMA points out that in creating a strategy around data mastery, insurers need to take an incremental approach.
"If you try to sort through all the potential data sources out there, how you might combine and analyze those is overwhelming," says Breading. "Most insurers are evolving from their current base. They have a current set of proprietary information and a current set of external data sources. They build on those one by one, finding ways to augment the data the already have and the analyses they can already perform."
Lockton has what VanOpdorp calls a "fluid" strategy around data where the sources of information and technological approaches to analyses are secondary to achieving the outcomes desired.
"We tend to get caught in the noise because there's so much information. It's easy to get distracted by a lot of the charts and graphs out there, and it causes overload. The challenge is to sift through what's available and be sure it helps us achieve our goal, whether it's finding ways to make the workplace safer, reducing costs, or improving the overall operation of clients and prospects. We are always focused on honing in on something actionable, something that is truly driving the outcomes for our clients."
In a recent project, Lockton helped a client assess its return-to-work program for injured employees. Despite putting in place a number of incentives designed to encourage employees to get back on the job, the client actually found lost workdays increasing.
In addition to analyzing structured claims data, Lockton pulled together information from claims notes and other unstructured sources, using QlikView to mine the data, test assumptions and visualize results. From its analyses, Lockton discovered that what the business was doing was actually counterproductive to its return-to-work goals.
"All the client's efforts were around incentives to get people to return to work," VanOpdorp says. "From a deep dive into claim notes, we found that, in their zeal to reduce the number of lost work days, they were actually encouraging injured workers to come back too soon. As a result they ended up being out longer due to re-injury."
VanOpdorp says that "understanding the 'why'" in the analytic process helped Lockton identify a course of action that made sense for the client. The potential impact could equal millions of dollars in terms of reduced claims costs, as well as increased employee wellness.
"As a result of the analysis, the client has really started to focus on the other pieces of the claims puzzle," says VanOpdorp. "It's not just about returning people to work, it's about ensuring that they are looking after the health and welfare of their employees."
Visualize Results
"SAS allows analysts to uncover things in the data that they wouldn't see on its surface, and to see it more graphically," Wilson says. "Using visualization tools on top of data gives business users an easier way to interact with it. That's an additional value on top of the predictive models we are able to build."
Visualization is important in a big data environment. "It's not producing spreadsheet after spreadsheet with lots of numbers," say Stuart Rose, global insurance marketing principal for SAS. "[Insurers] can use geospatial information and bubble charts, identify trends, and drill down into information."
For instance, another SAS insurance client found that, a year after a major storm, new claims were being reported on properties that had been previously repaired.
"Once they [the client] drilled into the data, they could see that claims were related to one or two particular contractors that came from 50 miles away. That led them to a deeper investigation that found those contractors were targeting the region, going to insureds who were more senior and convincing them to make new claims. You wouldn't be able to see that if it were just a spreadsheet—you needed to visualize it, to see it on a map."
As more and more data becomes available to insurers, they also to stay focused on the 'why' in order to turn the flood of information into more than just a trickle of knowledge.
"Any kind of investment [in big data] will have to do one of two things: increase revenue, or make more profit," says Anderson. "Many carriers have targeted reducing claims costs and uncovering fraud, but there is the opportunity for greater application and advances throughout the business."
"When we take data and transform it into wisdom, that's what it's about," VanOpdorp says. "That drives a deeper relationship between us, the carrier, and the client, creates a more substantive discussion around issues, and brings the parties together to solve real business problems."
© Touchpoint Markets, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.