The Elusive Goal of Enterprise Data Warehousing

This Month's Experts

BARBARA KOSTERCIO for individual life insurance, Prudential (Newark, NJ, more than $371 billion in assets under management).

MICHAEL CONNOR Senior project manager, applied sciences, Zurich North America (Schaumburg, IL, $8.0 billion in annual premiums).

PATRICIA L. SAPORITO Senior insurance industry consultant, Teradata Division of NCR Corp., (New York).

JOHN SEARCH Vice president, Millbrook, Inc. (Bethlehem, PA). ------------------------------------------

Q: Has data warehousing been successfully deployed in insurance?

A: Barbara Koster, Prudential: It has been at Prudential. The primary benefit is consolidation of customer information, allowing us "one complete view" of each customer. This helps us to better understand their needs and their relationship with us. Obstacles in the data warehousing process range from selecting proper warehouse size, to having a unified data model for consolidation, to larger issues, such as privacy and sharing customer information.

A: Michael Connor, Zurich North America: The technology has been deployed in insurance, but to date most projects have involved datamarts rather than enterprise data warehouses. The main benefits of warehousing are consistency of both the views of the information and the data content with associated business rules; the ability to easily tie information/or balance back to its source from one place; assuming proper architecture and implementation; cost reduction for information delivery; and data acquisition.

A: Patricia L. Saporito, Teradata/NCR: Companies that have planned their warehouses from an enterprise perspective but have implemented specific business-driven initiatives have seen significant value. Those that have tried to do it all at once have lost business support and funding. So have others that have taken the datamart route and later tried to integrate marts into an enterprise data warehouse. The most frequent mistake is failure to leverage data assets and facilitate use of the data throughout the organization.

We have seen a small measure of success in the deployment of specialized marts that facilitate focused queries. While not robust, these datamarts often provide relief for more immediate requirements. We see many failures result from trying to do too much in a single effort, rather than delivering incremental value within reasonable time frames.

Q: What technology issues need to be addressed to achieve enterprise-wide data warehousing?

A: Koster: Business intelligence tools, such as data warehousing, vary depending on the size of the warehouse your organization needs. Prudential chose a large warehouse—the IBM Universal DB2 database software running on IBM SP hardware—because we have massive amounts of client data. Some of the technology issues faced include the size of the warehouse, data quality and the amount of data. Lastly, you need to get plenty of intelligent people around you who know how to use the data efficiently and effectively.

A: Connor: Most of the main enterprise data warehousing issues are non-technical. Top level/senior management commitment and appropriate funding are essential. Business definitions and rules for the data must be agreed upon across the organization. Warehousing is a process, not a product, so one must understand all the components and their relationship to each other. Design for scalability from the start, or expect to do a major overhaul later on. Spend the minimum amount of time necessary for evaluating various vendor products.

A: Saporito: Enterprise data warehouses require enterprise funding and commitment; that means a corporate sponsor and use of a steering committee. Communication is key, as is business and IT alignment via a joint project team, managing user expectations and communicating value. Technically, a flexible data model, data transformation (80 percent of the effort), metadata to understand the data and appropriate analytical tools are musts.

A: Search: A dimensional model promoting flexibility and symmetry of design and approach is important for a successful installation. The successful warehouse project should be dynamic and ongoing. All attempts should be made to provide it with the required resiliency to change.

Q: What is the role of vendors and technology solutions providers?

A: Koster: Vendors can help by supplementing internal resources. They can provide you a "jump start" in establishing a data warehouse and building acceptance for its uses throughout the organization. It's important to ensure, however, that the knowledge around how the warehouse was built and its functionality needs to be successfully transitioned to the people in your own organization who will maintain it.

A: Connor: The clearest road to disaster is to let vendors/technology providers drive the solution—they need to fit your architecture, not the reverse. Most have solutions that are valuable in given circumstances, but you must ensure your situation is one of them. If anyone tells you they have a quick, easy answer to data warehousing, run away, hide your wallet and avoid future contact.

A: Saporito: Vendors can help accelerate development through proven frameworks, methodologies and tools, data transformation and project guidance. They can also provide logical and physical data models and reporting solutions that should require only 20 to 30 percent customization. It's important to understand use of a datamart vs. data warehouse. Datamarts should be built off the data warehouse. Further, insurers should beware of solutions that require them to change their business models or processes.

A: Search: Vendors have an obligation to introduce new technology solutions so that staff can focus on core business. To be successful, vendors must do a better job of understanding the client's needs and translating that to more customized solutions.

Q: What does an effective data warehouse support? What future technology developments could foster enterprise-wide data warehousing?

A: Koster: Data warehouses should support a number of functions, including information management, customer segmentation/analysis, marketing campaign development/analysis, list generation and strategic planning. Data warehouses should also support financial management, in that the data should be able to reconcile revenue vs. expenses, supporting the company's long-term marketing efforts.

A: Connor: Warehousing can support any reporting, decision support, data mining, etc., as long as the organization is committed to consistently using and leveraging this environment. Future trends to watch are a move toward globally based information; recognition that warehousing is essential to support e-commerce and Web-driven information needs; better integration, availability and practical application of metadata as an information value driver and cost reducer; and a paradigm shift away from front-end systems driving warehousing.

Peggy Bresnick Kendler has been a writer for 30 years. She has worked as an editor, publicist and school district technology coordinator. During the past decade, Bresnick Kendler has worked for UBM TechWeb on special financialservices technology-centered ... View Full Bio