(Bruce F. Broussard Jr. is vice president, businessdevelopment and strategic relationships with Insurity.)

The traditional application-centric approach to core legacyreplacement in insurance has long been a mix of modern systems,data conversion, and policy migration at time of renewal. Theadvanced configuration capabilities offered with modern systems hascertainly improved project success rates, but many of the majorchallenges and risks remain. Replacing the old application with thenew offers little long term architectural benefit since both shareessentially the same component and integration footprint.

Perhaps it's time to change the status quo. Rather thancontinue with the traditional application centric approach, a datafirst approach may be a better way to modernization; resulting inboth reduced risk and cost, and a future proof IT architecture.

The data first approach is not exactly new. Over the pastdecade a number of insurers have successfully leveraged this pathto modernization. Unfortunately, during that time data first hasonly been practical for larger insurers that enjoyed the luxury ofa deep pool of resources and the ability to take on complex customdevelopment and long term maintenance of proprietary data andintegration structures. While these resource and complexityrealities have kept most insurers on the outside looking in,several developments over the past few years have changed theinsurance data landscape, most notably the maturity and acceptanceof industry standard ACORD data structures. As a result, thedata first approach is now within reach for insurers of allsizes.

With the application centric approach, the new core insuranceapplication is implemented to run in parallel with the legacy aspolicies are migrated from the old to the new. While policymigration at renewal minimizes complexity, this process may take amatter of years before all products have transitioned into the newenvironment. With the policy portfolio split across multipleapplications until the migration is completed, a range ofchallenges must be navigated, including how to produce consolidatedreporting and minimize duplicate maintenance efforts. Thebalancing act only gets more complicated as duplication is alsorequired for integration services to connect both the old and newwith downstream applications.

Opting for the data first approach eliminates this logisticalnightmare. Focusing effort toward integrating the data of thelegacy and new applications in a common repository, which I'llrefer to as the Core Insurance Data Component (CIDC), is theinitial step. The CIDC can subsequently be leveraged as thesingular source for reporting and integration throughout theduration of the modernization effort and long afterwards. Since theCIDC will always reflect the current and consolidated version ofthe data, a complete portfolio view is always available, regardlessof where any given policy is during the migrationprocess.

Many have attempted to build the equivalent ofa CIDC in the form of an enterprise data warehouse, independent ofa core modernization project. Unfortunately, a disproportionatelyhigh number of these data warehouse projects fail due to a varietyof challenges which typically include cost and resourcerequirements.

However, in the context of a modernization effort, most of thework required to build a CIDC is necessary to complete themodernization project. Since data mapping and integration arepart of any core modernization project, the only remaining step totake advantage of the data first approach is to create the CIDC.With the recent evolution in insurance industry data standards,three classes of options exist that are technically and financiallyviable for any insurer:

• Pure custom development involves the design,build, test, and implementation of the CIDC from scratch. This approach typically takes the longest to implement and is alsothe riskiest. Without a proven design to start the process, therisks of creating a repository that is incomplete or performingpoorly makes this an unattractive choice.
• Model accelerated development offers a viableoption for many insurers, and has been the favored approach by mostof the larger insurers who have successfully deliveredmodernization projects over the past decade. The design phase forthe repository is accelerated by using either proven vendor datamodels or a replication of the data model for the new applicationas the starting point; reducing a key risk of pure customdevelopment. However, the modernization schedule is stilllengthened since the repository must be designed, built, tested,and installed before the legacy and new applications can begin topopulate the CIDC.
• Vendor product implementation is a relativelynew option, as vendor products with readily implementable insurancerepositories have only recently been made available. Thisapproach is attractive since there is no design or developmenteffort involved, and the product license may be far less costlythan building the repository through either of the other twoapproaches. The vendor repository is already fully built, tested,and ready to be installed, avoiding any unnecessary delay to themodernization project. By leveraging industry standardintegration methods, such as ACORD XML, cost and time are reducedas well as the complexity of populating the repository.

By implementing and leveraging a CIDC through the data firstapproach the business and other downstream applications areinsulated from core application changes, both during themodernization project and long after. The CIDC's simplifiedintegration structure helps future proof the IT architecture,reducing total cost of ownership and reducing the ripple effect ofapplication changes across other components. The newenvironment provides a platform for standardized, balanced andreconciled data stores that lowers costs by eliminating processesrequired to aggregate and validate data from disparatesources. Also, the platform offers the opportunity to expandbusiness intelligence and analytics, better leverage Big Data, anddeploy a broader and more consistent user experience, includingself-service options.

Given the industry's disappointing results when it comesto application centric modernization and independent data warehouseprojects, there is no better time than the present to takeadvantage of an approach that kills two birds with one stone;delivering greater benefits at reduced risk. Change the status quoand start with the data first.