Analytics Emerge All around the Enterprise

narora.jpg

While the biblical injunction against envy is unequivocal, inside the walls of an insurance company a little envy may be a good thing, especially when a successful implementation of a technology in one department spurs its deployment elsewhere in the enterprise.

This case for coveting seems particularly strong in the instance of business analytics, which often enters an insurance company in one department only to spread elsewhere. While relatively established in claims and underwriting, analytics is also finding fertile ground in areas such as marketing and agency management. Insurers, it seems, have analytical needs across the enterprise.

Such was the case in Fond du Lac, Wisconsin, where Society Insurance purchased an analytics solution to help with underwriting, but is now beginning to deploy it in its claims and marketing departments, says Rick Parks, the company's SVP and COO. Parks relates that the company selected a solution from Denver-based Valen Technologies with this type of flexibility firmly in mind. "We were interested in having a solution that could transition into other roles in the company," he says.

Dax Craig, president and CEO of Valen, says he sees analytics spreading from traditional uses such as fraud detection and underwriting into areas such as premium audit.

However, Craig cautions carriers that just because an analytics solution worked well in one part of an enterprise, they shouldn't expect instant results elsewhere, especially if relevant data is wanting. "If you don't have the data, you don't have a model," he says. "We can only help solve a problem if we can get a statistically and actuarially defensible amount of data."

Parks also cautions that companies looking to spread analytics throughout the enterprise are in for some work. "We realized that this was going to be quite an investment in time, resources and funds," he says.

A NEW DEPARTMENT

Few people realize the effort necessary to spread analytics across the enterprise more than Neeraj Arora, director of Personal Lines Insight & Innovations, at Los Angeles-based Farmers Insurance Group. The department was founded two-and-one-half years ago to drive value to other parts of the organization. "At a very tactical level we are trying to spread analytics around the organization department by department, or even project by project," he says.

In an organization as big as Farmers, such an effort entails gathering, collating and make sense of billions of bits of data in order to create predictive models. One area Farmers is seeing results is in marketing, where it uses response rates from previous marketing efforts to populate a predictive model that calculates the likelihood a consumer will buy a policy based on a certain piece of mail sent to them. Thus, the models helps narrow down the mailing list, Arora says.

Carriers can further zero in on a target demographic by including other common predictive variables such as age, credit information and marital status, says Stuart Rose, marketing manager, Global Insurance, for Cary N.C.-based SAS Institute Inc. "Analytics can realize what those traits are in people who will need more insurance and tailor marketing efforts to them," he says.

Arora says the departments can now apply scientific rigor to decisions previously made on tradition or gut instinct. "Where statistical analysis can help you is by telling you if you're gut instincts are really true," he says.

Such insights may seem counterintuitive, and elicit incredulity from business users long accustomed to doing things a different way. Accordingly, proponents of analytics need to constantly judge the effectiveness of existing models, adjust where necessary and be wary of statistical anomalies. "We always have an independent test area to see if one model is truly working or if it's just picking up on an abnormality," Arora says.

THE HUMAN FACTOR

Not discounting the many technical issues facing widespread analytics implementations, the winning over of skeptical business users may be one of the thorniest ones. Arora uses a "show, don't tell" philosophy, where analytics gains a small foothold in a department, and can show business users, based on samples, the value of the models. "We have to continuously deliver results and prove that things are working," he says. "Our grand vision is to have not one department untouched by analytics, and to get business users to see value maximization potential both for the organization and the customer."

Society's Parks agrees that managing the people, not just the process, is important. "There's a learning curve for a company that adopts this sort of technology," he says. "You need to get operational people out there to understand what it does, how you can use it and why it's a good thing. Getting people who have operated a certain way, sometimes over a full career, to look at the world a little differently and to understand that these are tools that can help them be more effective, that's a big step there."

Valen's Craig says the only two options for insurers employing new analytic solutions are to retrain existing workers or bring in new people. "It's tough to do this in a traditional actuarial department," he says. "Actuaries have a significant role in any predictive analytics project, but they are not necessarily trained in this kind of work."

Rose also sees a more decentralized role for analytics, with business users becoming increasingly comfortable with analytics as the technology improves.

"Now, you don't need to be a statistician or a quantitative analyst to start building out predictive models," he says, stressing the importance of education. "Now there's the ability to go in there and get ad hoc reports easily. It's moving away from the actuarial and IT aspect, and business users now see a huge value in that and are pushing the boundaries."

Moreover, Rose says insurers need to cultivate analytic skills from within as the industry will be competing for the best analysts emerging from colleges.

DATA QUALITY

However formidable, the personnel challenges are not the only the impediments to the dispersion of analytics throughout the enterprise. Issues of data quality need to be addressed. Data culled from legacy systems can sully an otherwise pristine model. "Data quality is a huge issue," says Parks. "If it's not correct, you can make wrong decisions."

According to Parks, analytic initiatives do have a tendency to ferret out flaws in how insurers store their data. "One of the upsides to analytics is that the more you use it, the more you understand how good your data is," he says. "As you become a more robust data user, almost by definition your data gets better because you find the problems and fix them. When things don't look right, it forces you to dig a bit deeper."

Craig agrees that the importance of data quality can not be overstated, but says the perfect should not be the enemy of the good. "No one believes their data is good enough," he says. "But what we've found is that it is never as good or as bad as they think."

Indeed, carriers may have already done a lot of the groundwork for analytics in previous data warehousing initiatives. While those that invested data warehouses may come out ahead of those that have not, issues will emerge.

So, where in addition to established areas such as claims and underwriting, can a carrier use analytics to improve efficiency to gain competitive advantage? One of the emerging areas is agency management and effectiveness. For example, a carrier may want to know which agents are requesting quotes but writing few policies. "We're seeing demand from customers wanting to analyze production," says Tom DiMarco, VP of Technology and Operations for Alpharetta, Ga.-based iPartners LLC.

Rose says he, too, sees more insurers employing analytics to gauge agency experience. "Previously, a lot of the metrics were around revenue and retention rate, without looking at customer profitability," he says. "Now, with the ability to dive deeper and actually analyze the customer and their value and lifetime profitability, they can better value the efficiency of their agents. The data is predominantly there."

Another area where Rose sees analytics making inroads is in product development. Innovative product offerings such as pay-as-you-go auto insurance need a strong analytic underpinning, he says.

While it's only over the past two or three years that analytics have become ubiquitous, the market is evolving rapidly. Some credit this to a greater awareness among senior management at insurance companies of the value proposition of analytics. Others note that the interest in analytics has risen as the cost of data storage has come down.

Another obvious explanation is competitive pressure. In areas such as personal lines auto, analytics are seen as table stakes. It's not yet that way in many commercial lines, but may soon be. One consequence of this pressure is that companies want more information, more frequently, and want to distribute it to more people within the enterprise.

"What we're seeing is that monthly isn't good enough anymore," says DiMarco. "People want to see their data on a weekly basis - in some cases daily."

Lauren Belfiore, director of operations for Springfield, N.J.-based Riders Insurance Co., says the insurer gets monthly reports from iPartners, but is considering going from monthly to weekly. "With the market the way it is, you need to be on top of your data," she says. "I think daily is a bit of overkill, but everybody wants to analyze the data as quickly as possible."

Parks agrees that given the competitiveness of the marketplace, if carriers are not at the tip of the spear when it comes to analytics, they should at least be somewhere high up on the handle.

"Our take is that if you don't take advantage of the technology, you'll pretty quickly lose ground," he says.

For more about analytics, search "Assessing Business Intelligence" at www.insurancenetworking.com.

(c) 2009 Insurance Networking News and SourceMedia, Inc. All Rights Reserved.

http://www.insurancenetworking.com/ http://www.sourcemedia.com/

For reprint and licensing requests for this article, click here.
Analytics Data and information management Policy adminstration
MORE FROM DIGITAL INSURANCE