For a long time, the operational smarts of an insurance enterprise resided primarily within the collective consciousness of its employees. Over the past few years, the insurance industry has begun to invest in business analytic solutions that leverage this knowledge and seek to augment human decision-making with data-derived insights. In fact, the use of analytics by insurers is approaching critical mass.
To summarize briefly, the term "business analytics" is intentionally broad and encompasses two similar but distinct technologies. Historically, BI has become synonymous with operational metrics and monitoring, querying and reporting functions. Modern BI solutions generate scorecards, a collection of metrics and reports on a unified interface used to measure against objectives and dashboards, which display metrics in graphic form in order to give users a fast read of an organization's pulse.
While BI tools largely give insurers a high-level view of what's going on, predictive analytics promise insurers actionable knowledge and a granular view of their operations. Thus, if BI is a look in the rearview mirror, predictive analytics is the view out the windshield.
In the months ahead, the relationship between BI and predictive analytics will become increasingly intertwined, says Donald Light, a senior analyst at Boston-based Celent. "One of the emerging ways that analytics solutions provide value is as an engine that feeds information into dashboards or other easily digested reports," he says. "The question then becomes who is going to look at those reports."
Indeed, many of the issues carriers will confront as they increasingly incorporate advanced analytic applications are more cultural than technical. Across all industries only a small percentage of people within an enterprise use analytic applications (see chart). "By its nature, analytics can involve some advanced statistical techniques that are beyond the comprehension of the person in the average operational role," Light says. "Some analytics programs are making it easier for the average underwriter who has little or no training in statistical techniques to answer questions. There's been a lot of progress in usability for non-technical users."
John Lucker, principle, Deloitte Consulting LLC, says the increased emphasis on data visualization underscores the trend of analytics use spreading from a small cadre of quantitative analysts (quants) to a broader range of non-technical employees. Lucker says the dynamic is similar to Web design, where technically proficient coders initially had the tools to themselves, but eventually learned to split their workload with graphic artists to make sites visually appealing. "This is going to be very challenging for the quants because these problems require more of an artistic skill set," he says.
As carriers recognize the cultural challenges surrounding widespread use of analytics, Lucker says positions such as chief analytics officer will become more prominent within organizations.
Lucker also foresees analytics products becoming more task-specific. While earlier offerings were aimed at broad business processes such as underwriting, Lucker says vendors will now offer more targeted tools to address issues such as premium leakage in personal auto. "We're going to see more "widget-izing" of the solutions, where products are built specifically to solve some niche problem," he says.
However, this attempt by technology firms to address specific problems has come at a cost. "Some of these vendors that have tried to innovate based on venture capital are going to fail," Lucker says. "As a result, the products are going to disappear. This is something insurers need to be careful about because it can cause enormous disruption."
Another noteworthy trend, he says, is newly viable offerings based on R, an open source environment for statistical computing and graphics. Much as companies such as Red Hat helped turned Linux from a dorm room project to a big factor in data centers, vendors are now readying R for enterprise use. "It's a very sophisticated tool and now there are commercial versions of it available," Lucker says. "R is enormously popular with Linux people and giving other tools a run for their money."
Analytic solutions also are increasingly making use of geospatial offerings. This fits into a broader trend of carriers crafting "synthetic" data points to use in their models. Synthetic data is made by combining structured data, often from a third-party source, and unstructured data to create data points that previously did not exist. As an example, Lucker says a workers' compensation insurer could create a synthetic data point by combining address information from employee records with geospatial information to determine the average distance the employee traveled to a given worksite. The information might prove useful to underwriter because a relationship is known to exist between claim severity and the distance a worker must travel to get to work.
