Systems Integration: Is There a Better Way?

In an industry heavy with operational costs, improving efficiency has long been at the top of insurers' to-do lists. With the weak economy and increasing competition, streamlined operations are paramount.

To achieve the sleek operations that many insurers desire, carriers must integrate IT operations with technology. And to do this, they have to stretch their IT dollars further than ever. Besides rewriting legacy and direct links among systems—processes that can incur tremendous costs—carriers are turning to alternative forms of "systems integration," including creating data warehouses.

Simply uttering the term "systems integration" at an insurance carrier may make some IT and business folks cringe, as many carriers have been attempting to link disparate IT systems for years, even decades. With modern technology and practices, however, systems integration—normally associated with directly linking two or more systems together—doesn'talways have to be a gut-wrenching, drawn-out process that only yields marginal results.

But, given the challenges, why are carriers suddenly interested in consolidating systems, after a few years where they seemed to be focusing on other kinds of IT initiatives, including many Internet-focused projects?

"During the boom times, the tendency is to add technology that will help produce more revenue, get more clients and make money," says John Fry, managing director, financial services, at Enterprise Solutions Providers, Inc. (ESP), a New York-based systems integration and management consulting services provider. "When money and IT budgets become an issue, the quality of integration of all of the new technology comes into play. Many of the new technologies have been integrated with manual processes," or people re-keying information and transferring data. "The means to integrate the systems and the new technology were not really considered."

While it doesn't seem like a good business strategy to add new technology that is supposed to automate the business and then add staff to help it communicate with other systems, in many cases that is exactly what insurance companies did, relates Fry-mainly because there never is as tight a hold on the purse strings during flush economic times. "When the tide was in, they went out and built and bought a whole bunch of new ships," referring to carriers that implemented new technology and bought books of business from other companies, he adds. "Now that the tide is out, carriers are seeing a whole bunch of rocks, so they are deciding to do some dredging."

And the main reason for trawling the depths of the IT organization is to search for areas where costs can be cut and efficiencies gained. "Carriers are under immense pressure to drive cost out of the business," whether it is because of an insurer's recent IPO or due to simple business pressures, says Michael Aubin, chief operating officer at Metaserver, Inc., a New Haven, CT-based business process integration (BPI) services provider. "Insurance companies are now serious about reducing cost. Many insurance IT executives are asking, 'How does an insurance company differentiate itself?' The answer is by offering better customer service at a lower cost."

No ROI, No Deal

The directive for the search for lower operating costs is being passed down with an equally important directive that all projects and cost-cutting initiatives must demonstrate a return on investment (ROI)—something that generally was not asked for during the boom times, and that ultimately helped produce many of the disparate technologies that carriers are faced with integrating today.

Faced with tall ROI orders and short IT budgets, IT executives are discovering that integration doesn't have to be a multi-year, multi-million-dollar initiative, although "classic" systems integration-linking systems directly or consolidating legacy applications-is still taking place with much success.

For instance, when the chief marketing officer wanted better access to data for analytical purposes at Allstate Financial, a division of Allstate Insurance Co. ($112 billion in assets, Northbrook, IL), IT had to find a way to collect data from seven separate legacy systems—primarily back-end systems for processing annuities. "Many blocks of our business have been bought from other companies over time," says John Hershberger, vice president, database marketing, Allstate Financial. "Many of those blocks of business are on systems that have the same names, such as Continuum and Cybertek, and appear to be running the same software. But the systems have been so heavily modified to the specific needs of each business that they are unrecognizable.

"The cost associated with doing retrofits, or recoding" to consolidate the systems, "or even to map the information from one system to another, is not financially feasible," Hershberger adds. Since a strategy to consolidate the legacy systems would not be supported because of the high costs and the length of the project, Allstate Financial decided to build a data warehouse to house the information from the seven systems. The data warehouse, which is primarily used by marketers for demographic analysis, was running six months after its conception-a far shorter time than it would have taken for a complete systems consolidation or integration to take place, Hershberger asserts. Allstate Financial's data warehouse, which is used for decision support and analytics, not operational support (customer service), employs an Oracle (Redwood Shores, CA) database structure with SAS Institute's (Cary, NC) Enterprise Miner for data mining on a Sun Microsystems (Palo Alto, CA) e10000 three-terabyte production server. Data is extracted from the legacy systems with Ab Initio ETL (Extraction, Transformation, Loading) technology.

"We are not using the data warehouse for customer service," Hershberger says. "It is meant to improve business and is built to improve business-decision support." Originally developed for marketing, the marketing department only accounts for approximately 55 to 60 percent of the requests for information from the data warehouse. Access to the data warehouse has been expanded to other areas, and will eventually be available to distribution channels over the Internet.

But even though building a data warehouse or data mart may be an easier way to link information in systems together, it is not a slam dunk. "We've seen many cases of companies that have had three or four failures when it comes to building data warehouses," says Ron Barker, senior principal, and insurance area practice leader at Chicago-based Knightsbridge, a data warehousing solutions provider. "Many of the enlightened companies are taking another look at data warehousing, but they are taking it in smaller, bite-sized steps and relating all of the initiatives to business metrics."

One reason why data warehousing has such a tarnished reputation and was not really considered a viable IT alternative to systems integration is because of the numerous data warehouse failures that litter the IT landscape. However, Barker contends that most of those debacles were ill-conceived from the start. "Most initial data warehouse initiatives were leaps of faith," either by the business or IT sides, he says. "It usually happened when IT felt it knew the needs of the business better than the business executives. It seldom works out when the data warehouse is driven by IT folks. When the data warehouse is run by the business, it is generally successful."

But even successfully planned business-focused data warehouses can be costly, contends Metaserver's Aubin. "Depending on the focus, determining the volume of data that is needed and getting the information into a data warehouse is not an easy task," he says. "Sometimes it is not good to look at things from purely a data perspective. Carriers should look at the business drivers and the business process and let that drive the integration."

Determining the need for the integration of data from systems should be the first goal of any initiative, says Steven Landberg, managing director, Alpha Financial Services Consulting (New York). "Many companies take the approach, when looking to integrate data and systems, that they have to build a single system that will do everything," Landberg says. "And that is simply not the case."

Legacy's Legacy

According to Ron Young, general manager, insurance, at Siebel Systems (San Mateo, CA), "Having one single integrated system is very unrealistic. Many components do not work well together," he says. "The reality is that you will have 35-year-old legacy systems communicating with five-day-old systems," and that achieving that level of connectivity is dependent on models such as ACORD's (Pearl River, NY) XML Standards.

For instance, this all-in-one approach is especially common when companies are looking to integrate systems for customer service, or CRM. Landberg relates that he has seen many insurers making the mistake of trying to include every possible touch point in a CRM system, when in reality some of the functionality may never be used—like Web chat, a technology a few carriers have developed, only to find that less than one percent of its customers actually want to use the technology. "Take the five biggest pains"—possibly log jams in claims, manual processes in underwriting or key customer interactions that need smoothing—and start there," he advises.

Hartford Property & Casualty, a division of The Hartford ($167 billion in assets, Hartford), was comprised of five separate P&C companies before it moved to a single company model, says Kevin Yorgensen, assistant vice president, information services. "In the personal lines area, the five companies had grown the systems over the years," he says. "The systems were different and quite separate. That becomes an issue when it comes to working on things like CRM."

Hartford P&C needed to get a grasp on the policyholder information, but instead of physically integrating the separate systems from the five companies or falling into a trap of creating an enormous data warehouse for the entire P&C company, it focused on a specific business need. "The project was funded by the underwriting and actuarial areas," Yorgensen says. "Actuarial wanted to be able to look back over years of policy history in order to better price products and how to best select risks. In this case, the more data the better."

Greg MacSweeney is editorial director of InformationWeek Financial Services, whose brands include Wall Street & Technology, Bank Systems & Technology, Advanced Trading, and Insurance & Technology. View Full Bio

1 of 2