A data warehouse is a company-wide database that brings together data from disparate units or operations to support marketing and other decision-making in an organization. When the database is organized for a single business unit within an organization, it may be referred to as a “data mart.”

“Despite the high failure rates for data warehousing, people still want to do it,” said Seth Rachlin, co-founder, president and chief executive officer of New York-based Connect Systems Inc.

“Data warehousing can deliver significant return on investment across the enterprise,” said Mr. Rachlin. “This is true in many industries and it is particularly true in insurance. I still havent seen the insurance sector embrace these kinds of technologies the way industries like publishing and telecommunications have,” he added, “but theres certainly a lot of interest there.”

Mr. Rachlin also maintained that data warehousing technologies have matured significantly. “The technologies out there are pretty good,” he observed. “Its been out for awhile and people know how to do it.”

He added that recent development of Web-hosted technologies have made data warehousing accessible to more end users. “From a technology standpoint,” said Mr. Rachlin, “implementing a data warehouse is not that difficult.”

The principal reason so many data warehousing projects still fail, according to Mr. Rachlin, is that “companies are not good at running projects at an enterprise level.”

“Particularly within the insurance sector, there is a product or functional focus versus and enterprise perspective,” he continued. “They understand everything there is to know about a particular product theyre selling. They see themselves as completely tied to that product.”

Another reason for data warehouse development failures is that different business units within the same company will have different definitions of common terms, such as “premium” and “agent,” said Mr. Rachlin. This complicates the process of working with data from the various business units.

Executives can also get territorial when it comes to ownership of the data between business units, Mr. Rachlin noted. Claims of ownership might also extend to distribution networks and to the customers themselves.

Another consideration, according to Mr. Rachlin, is that while “data warehouses do great things, they also unsettle people and make them nervous.” The creation of a data warehouse can change the way a company typically does business, he noted, and as a result some shifts in corporate power might take place.

“It does threaten people,” said Mr. Rachlin of the data warehouse concept. He added that those who take on data warehouse projects “need to recognize that not everyone is your friend.”

Mr. Rachlin also pointed to the challenge of setting priorities in terms of what gets done first in a data warehousing project. Having a method and process to determine those priorities is “very critical,” he said.

What can be done to overcome some of these obstacles and meet the challenges? Mr. Rachlin said getting high-level commitment within an organization is a key factor, along with encouraging broad participation in the project within the organization. “You cant do this kind of thing on the stealth,” he observed.

One way of encouraging participation is to regularly communicate the status and progress of the data warehousing project to everyone concerned, said Mr. Rachlin. He pointed out that “cleaning up hundreds of years of data problems” and setting up the project takes time, energy and effort. This part of the project “doesnt appear to have immediate results,” so it is important to communicate frequently, he added.

Successfully developing a data warehouse involves “changing the culture, not just the systems,” said Mr. Rachlin. There is a need to “broker consensus” from all those involved, particularly in terms of “what things mean and how they will be reconciled,” he continued.

Getting consensus doesnt necessarily mean “someone wins,” Mr. Rachlin noted. “It may mean you have to tolerate and live with four different definitions of premium.”

Mr. Rachlin also recommended “federated ownership and governance” of data warehouse projects, rather than having one small group be in control.

Yet another way to help ensure success is to “tie funding to specific deliverables, and to specific lines of business,” said Mr. Rachlin. “Thats the surest way to make sure [budgets] are not cut.”

Another panelist, Mike Heilman, a business technology consultant at Nationwide Insurance Company in Columbus, Ohio, agreed on the importance of finding an “executive sponsor” for a data warehousing project. “If you dont get that commitment, you will not succeed,” he asserted.

Mr. Heilman also agreed that establishing data definitions is a key to success in data warehousing projects, as well as aligning the projects goals with the companys key business drivers, such as profitability.

In addition, Mr. Heilman emphasized that “project management skills are the single most important technology resource that comes to play on data warehouse development.” Such skills involve management of technology, people and processes, he explained. “You must balance all three. Any failure in an area means failure of the project.”

Reproduced from National Underwriter Property & Casualty/Risk & Benefits Management Edition, October 22, 2001. Copyright 2001 by The National Underwriter Company in the serial publication. All rights reserved.Copyright in this article as an independent work may be held by the author.

Contact Webmaster