Building a Better Warehouse.
When Conning & Co. asked a group of insurance companies in 1999 what they thought about data warehousing, only 20% of the respondents said they found data-warehouse systems to be effective to their company's operation, and 68% viewed the techniques as ineffective or not innovative. The outlook appeared bleak for data warehousing.
But in 2000, another study by Conning showed that insurers were beginning to resolve some of the challenges and recognize more of the benefits of data warehousing. The study linked data warehousing and data mining to success in nearly every aspect of a company's performance.
"Many companies have been very successful using data warehousing and data mining to do many different things, such as selling and pricing better, cross-selling, reducing expenditures and improving customer service," said Jack Gohsler, senior vice president at Conning, a research and asset-management firm focusing on the insurance industry. "However, there are many cases in which insurers have not been successful at all, spending a lot of money with little payback. Overall, there has been tremendous diversity in the results that insurers have reported."
The results of the 2000 study "Data Warehousing and Data Mining in the Insurance Industry--Floods of Information," encompassed across-the-board responses, ranging from optimistic experiences to those more pessimistic about their use of the technology. Conning believes the industry as a whole is slowly beginning to move into the data-warehousing arena.
A Historical Perspective
In the 1960s, data collection was a large part of companies' practices of accumulating client information. With the development of computers, companies relied on data collection to evaluate total revenue and compile customer demographics. In the 1980s, the enabling technologies of relational databases and computer languages such as ODBC brought forth the onset of data access. It was during the 1990s and the time when Bill Inmon--deemed the "father of data warehousing"-- published his first book on the concept that companies began to use data warehousing to consolidate information from disparate operational systems into one source for reliable and accessible information. Data warehousing is a generic term used for the system for storing, retrieving and managing large amounts of any type of data.
"While data warehousing has been in existence for eight to 10 years or so, I think there has been a much-increased focus on it over the last two to three years," Gohsler said.
As a method to improve the sharing and storing of client information in data warehouses, data mining is now beginning to make advances into these systems. Data mining, which relies on advanced algorithms, multiprocessor computers and massive databases for information storage, is described as the extraction of hidden predictive information from large databases or, more simply, the "mining" of data placed inside warehouse systems. Data mining tools are used to predict future trends and behaviors, allowing companies to make proactive, knowledge-driven decisions.
Success for Some
Despite the diversity of opinions about the new technology, several insurance companies are finding success with their data-warehousing systems and their related data-mining tools. In fact, many of these companies believe the techniques are now slowly beginning to find a niche within the insurance society.
Many of these companies believe data warehousing boils down to one thing: the company's survival. "You can't possibly analyze data at this level with these kinds of volumes, unless you are taking this approach," said Sharon Sibigtroth, managing director of strategic data technology at Axa Financial, based in New York City. After 10 years of operation with a data-warehousing system, Axa is finding it to be invaluable and crucial to the organization's operation.
"Although we had much value from it over the years, we are really seeing tremendous value now that we are getting into the [customer relationship management] space, e-business and analytics around using the data with data-mining tools," Sibigtroth said. "I think having a data warehouse in place for us has really been advantageous, because we have a lot of data pulled together and don't have to spend time doing it, which is a very time-consuming, tedious and costly effort."
Chubb Corp., Warren, N.J., is another company that has recognized the benefits that data warehousing can bring to its operation. The company built its system in July 2000 and unveiled it in November. Chubb executives are hopeful that the company's data warehouses and data marts, which are data structures designed to facilitate end-user data analysis, will be successful in producing considerable savings.
While some insurance companies are beginning to embrace data warehousing, some companies in the Conning study cited poor data quality and lack of understanding of the data as two of the chief reasons for its failure.
According to Conning, poor data quality is responsible for many of the time and cost overruns insurers have encountered in building a warehouse. The study indicated that these difficulties are most often the result of insufficient or inappropriate edits and validations of data in prior years when policy and claim transactions were processed.
The issue that most often impairs data-warehousing efforts is data quality, Conning's Gohsler said. One of the things companies want to do to make data warehousing's benefits more immediate is to input several years of historical data and use it immediately to identify trends. Data warehouses really allow companies to look at the data more closely than in the past. "If you load data that hasn't been edited very carefully and has not been consistently defined, the data is of questionable quality. Then when you try to use the data more rigorously, you find that you don't have the level of quality you expected," Gohsler said. "Therefore, it is almost like the old accounting term 'garbage in, garbage out.'"
Cleaning the Data
One of the ways several companies in the Conning study tried to overcome poor data quality was by loading raw data into the system and trying to clean it after the fact. In many cases, this approach resulted in the damaged credibility of the companies' data-warehouse system. But the study found that the companies that cleaned their data before loading it into a data warehouse encountered fewer problems.
"The heavy lifting involved in this whole process is just getting the data moved from source systems, cleansed, consolidated and making it available to the end users. By comparison, choosing and implementing tools that you have to put in front of the end user for them to slice and dice data is the easy part. The hard part is getting the data in order and making sure it is of high quality," said Jeff Hoffman, vice president of business intelligence at Chubb.
Some insurers also have problems understanding the data in a data warehouse. "You have to focus on marrying a detailed understanding of the data and how to organize it with an in-depth understanding of the business. Companies must be able to do this if they expect to be able to use the warehouse for reasons that they may not even now understand," Gohsler said.
Study respondents also cited the lack of universal accessibility to data being stored in the warehouse system as a challenge to users. According to the study, information remains underused without universal access to the data warehouse.
One of the key challenges is integrating tons of data in the mainframe data warehouse with business-specific data in data marts on the open systems client-server platform, said Kent Bauer, director of data mining and analytics at Axa Financial. "We are also pulling in from Axa Financial's Web portals. So the ability to integrate all that data and maintain a customer and household focus are key challenges," he said. "In the past, you spent 70% to 80% of your time gathering and cleansing data, and only 20% to 30% of your time analyzing it." Bauer believes the way to remedy this challenge is to shift that paradigm. Once accomplished, the benefits of teaming data warehousing with data mining will yield an enhanced customer focus and profits.
To promote universal accessibility to data in existing systems, companies may need to team executives and the information technology department. "These individuals can bring resources and technology insights together to result in the development of an effective system," Gohsler said.
There are several other potential challenges data-warehouse users or those interested in setting up systems may face. "One of the things that becomes painfully evident, for all the above reasons, is that many data-warehouse initiatives take longer than planned and cost more," Gohsler said.
With initiatives that are potentially impaired by people and organizational issues (e.g., potentially bad historical data), time and cost overruns are not unexpected, he said. "These initiatives require technology and business people to work together in ways that are much different than in the past." Gohsler believes that companies are becoming much more familiar with data warehouses and their benefits and are becoming better at dealing with the potential challenges, increasing their likelihood of success.
Making It Work
One way companies can overcome challenges and garner success is by recognizing the benefits. "This can be accomplished once they realize the system is helping them make a marketing transformation ," said Richard Harvey, financial-services executive for IBM Business Intelligence Consulting. Companies will have success because data warehousing provides them with the ability to begin to understand their market, do segmentation and understand their customers from not just basic demographics but from their behavior. It allows them to get a feel for who their customers are and how long their best customers are staying with them. This includes knowing the potential for those best customers to buy additional products from the company so the company can make sure it is investing in the right kind of products to attract and retain these customers, Harvey said.
The Conning study points to several other benefits an effective data warehouse can bring to insurers: the ability to identify and anticipate customers' and potential clients' needs, understand and discover hidden exposures, and develop protocols for loss control and improving treatment outcomes. Companies undertaking data-warehouse projects need to understand their existing customers and the characteristics of their best, and most profitable, clients.
The Conning study offers an example of a large property/casualty insurer achieving success with its system. The company, which Conning did not name, has been enhancing its data-warehousing capability since 1993 and achieved a $3 million to $4 million per year cost savings for data acquisition through the elimination of redundant data requests. The financial savings resulted from the company's reliance on its data warehouse to empower its agents with desktop underwriting capability and prequalified leads. In addition to producing much financial success, data warehousing is increasing the company's capability to provide its agents instant access to applicants' information, including credit reports, driving records, claims history and family relationships, the report said.
For those companies not yet benefiting from their data-management effort, Gohsler believes success can be achieved once companies learn how to harness its power. "Companies need to recognize the extent of people and organizational changes that the data warehouse will cause and proactively manage changes," Gohsler said. In addition, he believes companies should pay particular attention to data-cleansing requirements and allocate sufficient time to complete the process.
Successful implementation will have rewards. "Companies will have data definitions that are consistent. You'll have a data warehouse that is both scalable and reusable. In addition, it will bring a one-stop shopping arena to do analysis separate from the operational world. You can have separate analytical processes by the business going against their appropriate business data marts ," said Craig Kovar, team leader of the enterprise data warehouse at North-brook, Ill.-based Allstate Insurance Co. Allstate created a data mart for its claims business requirements in February 1988.
"It is also imperative for companies to develop for results--not just analysis--and seek immediate results while planning for the future," Gohsler said. The Conning study also recommends that companies encourage as many employees as possible to use the data warehouse. This means implementing training programs. Companies also should commit to ongoing maintenance and enhancement of the warehouse to achieve success.
A Look Ahead
The future of new data technology is brighter than some may expect, said Chubb's Hoffman. "I think the industry has really started to embrace [data warehousing]. If you were to look out across the landscape 18 months ago, you would probably see that we were well behind the other financial-services companies, such as the banking and investment sectors, in this area," he said. "They may be ahead of us now with some other things, but I think the insurance industry has caught up and a lot of companies are now implementing data warehousing and data mining and trying to embed it into their processes." Chubb believes that as it engages more business units within the company, its opportunities to leverage information across and throughout the organization will grow. The company is relying on data warehousing and data-mining techniques to assist in this effort, Hoffman said.
Since data warehousing and data mining are still being introduced throughout the insurance industry, their long-term effects may not yet be clear. The Conning study predicted, however, that data warehouses and data mining would continue to overcome their developmental problems and, as major issues like data cleaning are resolved, more extensive initiatives will be undertaken.
"Overall, we see tremendous value. The Gramm-Leach-Bliley Act is probably going to spawn increased interest in more enterprisewide data warehousing. And people are becoming more experienced with it. There have been more successes and people are much more comfortable with its use. In addition, technology is far improved and less expensive," Gohsler said. For example, there has been much improvement in the processes supporting data warehousing, such as data modeling, and in processes to extract, transform and load data.
Once companies overcome challenges and recognize the positive impact data warehousing and data mining can bring to the operation, they may achieve their biggest benefit: "Understanding your customer... it's that simple," Axa Financial's Sibigtroth said.
Data-Warehouse Savings Outweigh Costs
While the data-warehousing concept is garnering interest from some insurance companies, a 2000 Conning & Co. study reported that several companies' projects failed to deliver the benefits expected of them and proved to be expensive to develop and maintain.
In 1998, the average user spent $6 million on its data warehouse, according to a study by the Meta Group, a leading research and consulting firm that focuses on information technology and business transformation strategies. Insurance companies spent $5.5 million on their data warehouses in 1998, as compared to an average of $1.9 million in 1996, the study said. The Gartner Group, a technology consulting firm, also looked at similar data. In a survey of 2,000 companies that launched data-warehouse projects, the Gartner Group estimated that U.S. companies spent $7 billion in 1999 on the creation and operation of data warehouses. In addition, the firm estimated that the cost spent on these techniques has grown by 35% annually since 1996.
Financial difficulties often arise because companies are unaware of all the costs associated with setting up and starting the process. Such costs include the purchase of hardware and software and fees associated with the data-cleaning process. The cost of extracting, cleaning and integrating data represents between 60% and 80% of the total cost of a typical data-warehousing project. But many current users believe the savings.
may outweigh the costs. Because the systems eliminate the need for information technology staff to capture, clean and access the company s disparate data-a typically large expense to companies--financial savings may result. According to the Conning study, several companies indicated that in the absence of a centralized data warehouse, between 60% and 80% of their information technology department's expenditures for business analysis support involved data discovery and cleaning activities. A data warehouse can eliminate the need for this process and drastically reduce or even eliminate these associated costs. Cost is beginning to take a back seat in insurers' decisions about whether to develop a data-warehouse system. "Price structure has really come down over the past five years," said Sharon Sibigtroth, managing director of strategic data technology at Axa Financial.
Gramm-Leach-Bliley Creates Greater Need for Data
The Gramm-Leach-Bliley Financial Services Modernization Act of 1999 has contributed to the use of data warehousing and data-mining techniques in both the banking and insurance industries. The act opened the door for these two markets to enter one another's business. As a result, both industries were compelled to learn more about their customers so they could cross-sell investment and insurance services. Data warehousing has become an effective way to accomplish this task.
"[Gramm-Leach-Bliley] is really consistent with an expanded view of the customer," said Jack Gohsler, senior vice president at Conning & Co., an asset-management and research firm. As companies begin looking at customers more holistically and want to compete for a larger share of their business, one of the things they are going to need to do is get more data and become much more customer-centric. Consequently, Conning thinks that the scope of data warehousing is going to increase, Gohsler said. "We are going to have to know our customers and distributors better. Data warehouses can probably drive a lot more competition advantage with GLBA than [they] could before the act was in existence," he said.
One way Gramm-Leach-Bliley may prove beneficial to insurers is by increasing their opportunities to partner, both within their own organization and with other financial-services companies. According to a 2000 Conning study on data warehousing and data mining in the insurance industry, "Companies will need enterprise data warehouses both to assess the potential value of these partnerships and to enable the smooth transfer of data that will be an important determinant of the success of these efforts."
Tied in with the benefits of Gramm-Leach-Bliley, several companies already are finding value in the use of data warehousing to cross-sell products. "We have a number of different marketing efforts we are tying it to," said Jeff Hoffman, vice president of business intelligence at Chubb Corp., regarding Chubb's use of its system in the cross-selling arena. "We are using the technology to provide our staff with more customer insight. These tools can help our agents and brokers by presenting them with information that will focus their marketing efforts. Instead of trying to cast a wide net on a whole lot of customers, the tools allow us to give them information that can target specific groups of customers, producing a higher hit ratio," Hoffman said.
Currently, both State Farm and Axa Financial are in the early stages of implementing their data-warehousing systems into cross-selling programs.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||challenges of data warehousing and data mining in insurance industry|
|Comment:||Building a Better Warehouse.(challenges of data warehousing and data mining in insurance industry)|
|Article Type:||Industry Overview|
|Date:||Mar 1, 2001|
|Previous Article:||Automation Advantage.|
|Next Article:||Capitalizing on Change.|