site_id
Enterprise data warehousing (EDW) is experiencing a renaissance of sorts. Although it was explored and practiced throughout the 1990s and early years of the 21st century, EDW has recently evolved from a highly theoretical approach into a mainstream architecture. The design and implementation of EDW in the early days presented challenges that were exacerbated by lack of appropriate hardware architecture and acceptable performance, as well as technologies that aimed to facilitate the integration of various data sources into one logical and physical model but were too immature to achieve it. Today, armed with advanced tools aimed specifically at building large enterprise-scale data repositories, more and more organizations are trying to achieve success in EDW within a short period of time.
In addition to the more avant-garde, information-centric businesses, such as telecom and finance, EDW is now gaining greater acceptance in traditionally conservative (from an analytical methods point of view) industries such as insurance, government operations and pharmaceuticals. Today, there is greater emphasis on performance management in government, predictive analytics and financial governance in the insurance industry, and sales force alignment in pharmaceuticals. These activities and others require information that spans multiple business lines and subject areas, a wider time horizon and a larger-than-ever number of participating data sources.
The availability of high-performance hardware and software infrastructure components has influenced this late bloom and made EDW an achievable goal even for smaller organizations. Technologies have become available for extraction, transformation and loading, relational database management systems and business intelligence platforms, thus removing two of the greatest barriers EDW faced in the mid-nineties: poor implementation of the architecture and sluggish processes of information delivery. Failure of popular architectures - such as pure bus architecture for large integrated data systems - has also driven an upsurge in EDW initiatives. These factors, combined with the desire to build bigger and better structures to support more-complex business intelligence and operational applications, have injected new life into the EDW school of thought and driven major undertakings currently underway in many organizations.
Insurance businesses, particularly the mid-sized companies that started their data warehousing efforts in the early 2000s, are showing a strong interest in EDW. They see this strategy as a means to wrap their business processes with information-rich solutions generated by integrated data warehouses where a multitude of sources are combined under one logical structure. As is always the case with good but rather complex products, they do not always turn out as planned, especially in their early releases.
Insurance organizations need to make sure the proper approach is taken in order to ensure the successful design and implementation of an EDW.
1. Integrate, integrate, integrate
At its most granular level, an insurance organization revolves around the policy and several key business functions: underwriting, claims, actuarial and finance/investments. Over the years, the legacy business has created numerous silos of data stores, each with their own data quality protocols, business rules, data structures, etc. Each silo profile usually reflects one line of business, or a major component of the business, such as claims or policy administration. In some extreme cases, no two silos are alike in terms of structures, data types and business rules, and, as a consequence, neither is the information they produce. In order to create an environment conducive to enterprise-level business intelligence, it is necessary to integrate all these types of data into one logical and physical structure that reflects the business architecture and process flow.
It is only then, when a high degree of integration is achieved, that a comprehensive view of the insurance business can be created.
2. A multi-polar world of data - Integrated subject areas and data domains
In order to unify data under a single architecture, and supply it to the business for meaningful business intelligence, all transaction systems or information subject areas should be represented in the warehouse. Complete, clean data from all business areas will enable the creation of a unified and unique view of the entire organization's business processes and their consequences in the market place, via basic unaltered data -- or, as it is commonly known, a "single version of the truth." The multiple subjects allow "information catering" for anyone interested in extracting intelligence from the data. Simultaneously, the integrated subject areas provide a complete view of the enterprise based on an unaltered and unique value of the objective business facts collected by the transaction systems.
Creating a complete view of the business is a long journey, requiring a road map that clearly identifies all milestones in building the warehouse. This will enable the integration of various subject areas in an iterative fashion based on company goals and objectives. This process of integrating subject areas will be threaded by the same architectural design concept and, like pieces of a puzzle, will add to the details of the full business picture.
It is also important to note that special consideration should be given to both logical and physical designs. Logical integration does not necessarily mean a monolithic physical architecture. The physical architecture should be flexible and allow for quick delivery and superior performance.
Tweet This