Business Intelligence Best Practices - BI-BestPractices.com

Collaboration. Communication. Community.

 
 
 Printer-friendly
 E-mail to friend
ADVERTISEMENT
The Western and Southern Life Insurance Company: A Data Warehousing Success Story

by Mark Kreyenhagen, Connie B. Robbins, Dr. Elaine Crable, Dr. Mark N. Frolick
In the summer of 1998, the Western-Southern Financial Group embarked on a project that would change forever how it viewed its customers and made strategic business decisions.

In the summer of 1998, the Western-Southern Financial Group embarked on a project that would change forever how it viewed its customers and made strategic business decisions. In August of that year, senior management approved a five year project appropriation to build a state-of-the-art data warehouse. While the marketing team was the primary sponsor and business driver, other groups saw the benefits of a real-time decision support system.

ford_logo

Through 2003, the project generated $30.5 million in estimated benefits; implementation cost $4.2 million. The costs were fully allocated, meaning that infrastructure development, software purchases, annual maintenance, marketing program costs, and salaries were all accounted for. The resulting ROI of 614 percent far exceeded management’s original return goals.

Introduction
Like many industries in the 20th century, the life insurance industry has seen a plethora of changes (Chesbrough, 1999). The industry has gone from being heavily regulated to being very market driven. Much of this change is due to the repeal of the Banking Act of 1933, commonly known as the Glass-Steagel Act. The result of this repeal: lines between the insurance and banking industries blurred. Historically, the life insurance industry had been based on what is known as a home-collect sales philosophy, unique to Western-Southern in the early 1900s, and allowed lower- and middle-income families the opportunity to purchase affordable life insurance. Agents would come to their homes weekly and collect a small amount of premium to keep the policy in force. In some cases, the premiums could be as little as five to ten cents per week. The home-collect philosophy provides many cross-sell and up-sell opportunities, as well as referrals for new business. Home-collect sales, however, also create an environment that can hamper the organization’s ongoing data collection of current clients. Sales representatives typically carried customer information in their heads and often times had no means of communicating this information to the home office. Invariably, the bulk of the information that is known about clients left with the sales representatives when they left the company. As one might well imagine, the information that these departing sales representatives took with them could be highly valuable to life insurance companies.

Like many insurance and financial services companies, the basic sales philosophy of face-to-face contact with customers has been a cornerstone of Western and Southern Life’s history.

In addition to the issue of customer information not making it back to the company from sales representatives, there have been other areas that prevented Western and Southern from fully capitalizing on the strength of their customer base. These issues include the segmentation of data into various legacy systems, not being able to effectively determine which customers are high value customers, and not being able to obtain a global view of the various products that customers own.

In order to better capitalize on this information and become more customer driven, Western and Southern decided to create a marketing data warehouse. This paper will provide an overview of Western and Southern Life, discuss the process of requirements and tool selection, and examine the various data warehousing campaigns undertaken at Western and Southern.

Western and Southern Life
The Western and Southern Life Insurance Company was founded in 1888 to provide whole life and term life insurance to the moderate and middle-income market. In the 1980s, this Cincinnati-based company diversified into a financial services organization. In 1998, Western and Southern Life became one of seven companies under the corporate umbrella of the Western and Southern Enterprise, which provides life insurance, annuities, mutual funds, asset management and other financial-related services to millions of customers nationwide. At that time, the enterprise had $13.9 billion in total consolidated assets, and was licensed in 40 states, with 200 sales offices.

In 2001, the organization adopted a new name: Western & Southern Financial Group. As a mutual holding company, the innovative service-oriented family of companies became a nationally recognized leader in consumer and business financial services. The company went through a period of rapid growth and by the end of 2002, assets owned and under management exceeded $28 billion.

With this rapid growth came the need to further refine Western & Southern Financials’ vision for the future of the organization. They defined the vision as:

 

to continue our evolution into a financial services powerhouse which capitalizes on our position of incredible financial strength in order to be a well-recognized, world-class (and best-in-class) financial services organization. We will provide superior value to all stakeholders and be a high-performance enterprise, which is a true leader in every sense of the word.

To implement this vision, Western and Southern decided to focus on the strategic business processes that were the core of the business. These processes included customer retention, customer service, cross-selling, lead generation, target marketing, and product generation. The vision proved to be the driving force behind the creation of the marketing data warehouse (Gaskin, 1998).

The Data Warehousing Initiative
One of the core functions of Western and Southern Life is the marketing process, since it is the primary point of contact with the customer. With the reorganization of Western and Southern Life, the marketing group was charged with three goals:

  • Become customer driven
  • Earn customer control via top-flight customer care and partner with others who have done so
  • Build profitable relationships with consumers who have unmet needs for value-added financial solutions

To be successful, this strategy relies heavily on a deep understanding of the characteristics and historical behavior of Western and Southern Life’s current customers. A review of their existing analysis and reporting processes revealed that customer-related data was distributed throughout a number of production systems, and not readily available for analysis. In addition, Western and Southern Life could not ensure that customer data gathered by the sales representatives made it back to the production systems. As a result, information concerning customers was difficult to analyze for marketing research purposes.

A review of the insurance industry’s best practices revealed an increasing reliance on data warehouses to drive marketing strategies and customer relationship management strategies (Chesbrough, 1999; Conning, 2000). A marketing data warehouse was therefore viewed as a critical success factor for analyzing and predicting customer behavior at Western and Southern Life.

The fundamental benefit of the proposed marketing data warehouse was that it would provide a single, comprehensive, and detailed view of customers across business lines and business units. A data warehouse would become the nucleus for a new customer relationship management environment and be one that would ensure that the organization had a centralized repository of customer information (Parzinger and Frolick, 2001). Once a data warehouse was established, the creation of multiple integrated distribution channels could provide the convenience and consistency necessary to service consumers via field sales, the client relationship center, and the Internet (anywhere, any way, any time).

In addition, a data warehouse would be a way to build profitable relationships with consumers who had unmet needs for value-added financial solutions. The data warehouse would allow Western and Southern Life to become more consumer driven, a strategy that relies heavily on a deep understanding of the characteristics and historical behavior of its current client base (Moncla, 1998).

The data warehousing initiative was initially projected to take one year to build, cost $5 million, and achieve a fiveyear payback of $6.5 million. This payback would be achieved with incremental revenue/savings identified by Western and Southern Life.

As with all data warehousing initiatives, an integral element was involving employees with the right skill set (Marks and Frolick, 2001). It can be difficult to find the right combination of individuals with the marketing, analytical, technical, and management skills to make a project of this type successful. To be successful with this project, the organization needed individuals with a vision for the future and with the ability to deliver results fast enough to satisfy senior management’s concern for ROI. For these reasons, the data warehouse was rolled out in a number of phases with each phase addressing specific ROI measures. Prior to the development process, Western and Southern spent a considerable amount of time studying the data needed from legacy systems as well as examining the best practices in the data warehousing industry.

Platform and Data Selection
The first objective of the data warehousing initiative was to develop a corporatewide agreement on the major objectives of the data warehouse and to detail how the data warehouse would be utilized to address seven major corporate goals:

1. Better identification of “bad risks” that should not be sold policies at all

2. Further improvement to underwriting risk analysis by replacing the comprehensive life underwriting experience (CLUE) database with the data warehouse

3. Better identification and pursuit of additional life insurance sales opportunities within the existing policyholder base 4. Better prevention of lapse risks among current policyholders

5. Begin mining current policyholder data to improve potency of “outside leads”

6. Identify characteristics of prospects who are not converted to customers

7. Improve on the recruiting process by creating a custom database for agent applications and applying mining techniques to that database

Once their issues were identified, information requirements for the new system were evaluated. This included the analysis of more than 2,500 data elements in 25 legacy systems.

In addition, the industry’s prime vendors in the areas of online analysis, data cleansing, data appending, data mining, and campaign management were evaluated. Table 1 lists the data warehousing tools finally used in the data warehousing development process.

Data Warehousing Tools Used
Online Analysis COGNOS: OLAP
Impromptu: business context reportsPowerPlay
Data Cleansing Trillium: data cleansing and householding
Data Append ACXIOM: CDI tool
ACXIOM: InfoBase
Data Mining and Campaign Management SAS Enterprise Miner: data mining
TransUnion: credit scoring
Table 1. Data Warehousing Tools. Once information requirements for the data warehouse were determined and the proper tools selected, the data warehousing development process was initiated.

Data Warehouse Solutions
The director of database marketing took an extraordinarily aggressive approach to establish the data warehouse as the nucleus for Western and Southern Life’s initiatives. Once the business partners (see Table 1) and the project teams were identified, a regimented plan was rigorously implemented to have the data warehouse in operation within one year. This plan would provide the ability to gain an immediate ROI on some of the first marketing campaigns which would provide the positive momentum and necessary senior management support necessary to continue the data warehousing efforts for this 113-yearold company. The success of the initiatives (campaigns) would test the capability of the data warehouse initiative (Moss and Adelman, 1999).

Campaign No. 1 (Persistency Campaign)
The goal of the first data warehouse campaign was to implement electronic payments as the preferred method of payment. Using COGNOS Impromptu, it was determined that policies sold on the Electronic Funds Transfer (EFT) payment mode were more persistent than those policies sold on mail payment modes. (In the insurance industry, persistency refers to policy retention, or longevity, over time.)

The campaign was tested in a pilot program from February 1 to April 30, 2000, before full-field implementation in July 2000. The pilot program cost $17,000 and returned $77,000. The full-field rollout cost an additional $5,000 and the net present value (NPV) of profits from this initiative was $4.2 million. This benefit continues to be realized annually, as new buyers who would have previously bought their policies on a monthly bill mode now choose the electronic option. Western and Southern Life today has approximately 55 percent of their monthly payers on electronic payment versus 35 percent before this initiative began.

The second phase of this campaign converted existing mail-pay and home-collect business to the electronic payment model. The data warehouse was used to prove empirically that electronic payers were more than twice as persistent as monthly payers. The results of the project also demonstrated to management that there were pockets within the company that were effectively selling electronic payments but that overall the company was still trapped in the home collect and monthly bill procedures that had been employed for decades. Changing this culture required hard data for monitoring the shift. The data warehouse provided sales management with the information it needed to correct this payment imbalance. As a result, electronic payments increased from 35 percent to 60 percent within three years. This campaign was piloted from June 12 to August 31, 2000, and implemented full field in October 2000. The cost of the pilot program was $2,000; it returned $154,000. The full-field rollout cost was $103,000 and had an NPV of $3.9 million. This campaign was repeated from September 3 to December 31, 2001 at a cost $103,000, and yielded an NPV of $1.6 million.

Campaign No. 2 (Cross-Sell Campaign)
The first cross-sell campaign was an annuity initiative. The data warehouse was used to identify prospective annuity purchasers. A profile of current annuity purchasers was developed, and the same characteristics were identified in customers who did not own annuities. The field sales agents were provided lists of clients who fit the annuity-owner profile. This campaign was piloted from January 17 to April 30, 2000, prior to full-field rollout on August 14, 2000. The cost of the pilot program was $11,000, with a return of $55,000. The full-field rollout cost an additional $26,000 and had a return of $861,000.

This campaign began to show the field sales reps the power of targeted, segmented marketing. Before using data mining, it was very common for most marketing campaigns to see anywhere from a one to three percent response rate. This campaign generated a substantial response rate of 8 to 10 percent. This campaign has been so successful that other similar cross-selling campaigns had been initiated using the same focused approach. Agents benefited from this campaign with higher sales; customers benefited by only being contacted if they appeared to be likely buyers; and marketing benefited by marketing to a smaller portion of the customer base.

Campaign No. 3 (Up-Sell Campaign)
The next opportunity involved identifying clients who were deemed to be underinsured. SAS Enterprise Miner software was used to identify customers who were underinsured and who fit a demographic profile of clients likely to purchase additional insurance coverage. The cost of the underinsured campaign was $75,000. The total net present value of the underinsured program was $525,000, with a return on investment of 596 percent. This campaign has been added to the marketing program and has been initiated on an annual basis with similar expectations.

To determine whether a customer was underinsured, the data warehouse team verified with LIMRA (an insurance research group) what amount of coverage that customer should carry. The team then scored the Western and Southern household base to determine whether they had this amount of coverage with Western and Southern. By using purchased demographic data, Western and Southern knew each household’s income amount. Many of the Western and Southern clients had reported in an earlier survey that the company was their primary carrier (and, in many cases, their only life insurance provider) so the team was not concerned about coverage held elsewhere.

The team then ran data mining models to determine which of the underinsured customers most resembled recent buyers. They scored the customer base and rankordered the households from “most likely to buy” to “least likely to buy.” The team turned over this scored list to the marketing department for campaign implementation. The campaign focused on the importance of life insurance and how most people need a multiple of their income for proper coverage.

Campaign No. 4 (Creation of Value/Potential Buying Index)
One objective of the data warehousing initiative was to determine how customers’ relationships were valued. Prior to the data warehousing initiative, no objective measurement existed that ranked customer relationships in terms of current profitability or future buying potential. Extensive work was conducted with the actuarial, sales, and legal groups to develop a customer relationship valuation measure that was not only correct but one that could be easily communicated to the sales force. After the model was validated internally, a prominent group of outside experts were brought into Western and Southern to review the calculations. Among the experts: IBM, LIMRA International, and Arthur Middleton Hughes (a respected CRM consultant and author). All reported favorable evaluations of the customer relationship valuation calculation.

In addition to the relationship valuation measure, the marketing department developed a relationship-building segmentation strategy that divided households into three groups (segments): gold, silver, and bronze. These groups were created using scores based on policy, customer, and household information. The scores examined current profitability and future buying potential. The segmentation also included a policy lapse probability score that indicated whether the relationship had a high, medium, or low probability of future policy lapse. The marketing team used this newly created index of gold, silver, and bronze to develop a contact strategy for each group, including sending newsletters, birthday cards, and holiday labels along with conducting annual reviews with each policyholder. Each group was handled differently according to the chance of responding to certain programs.

Later enhancements to this contact strategy included a lapse probability score, which assigns all households a score representing their likelihood of lapsing a future purchase. This score is calculated by examining more than 100 variables and is automatically re-scored each month using SAS statistical code. As relationships become more or less persistent, the data mining model identifies and recalculates the household’s probability of policy lapse.

Campaign No. 5 (Agent Prospecting)
The data warehouse team developed a state-of-the-art prospecting system designed to address three primary issues—agent recruiting, agent production, and agent retention.

This new agent prospecting system allowed field agents access to a data mart populated with prospects in their market area that would be likely to purchase insurance products because of life-changing events. Having a model to provide the agents with prospects helps to ensure that they are successful in their sales careers. The attrition rate of new agents is very high. New agents need to fill their days with productive activities. This system is one source of such productive activity. The data warehouse also provides credit-scoring models to agents so they can identify viable, likely-to-persist prospects. This prospecting application has been designed to be userfriendly, to be accessed over the company’s intranet, and to provide a range of demographic data elements that allow agents to custom filter a list of prospects.

As an offshoot of the new prospecting system, a data mining model was built to determine the characteristics of the company’s most successful agent. A particularly successful agent’s sales were analyzed for three years; this agent’s customers were more upscale, more stable, and between the ages of 30 and 60. To test the validity of the model, this agent was sent 2,500 prospects in ZIP codes where he did business. The agent was asked to identify which prospects he would pursue given basic household demographics. The agents’ answers were compared to the model’s predictions. The team determined that the model approximately 80 percent accurate. The model was further refined with another set of data and found to be 95 percent accurate.

The SAS code used to develop this model was loaded onto Acxiom’s server in Little Rock, and is now being used to select appropriate prospects for agents. This datamining model (known as Top Agent Model) has allowed Western and Southern to help all their agents model their best-practice agents.

Top Agent Model has been further refined. Customers are now purchasing polices that are 30 percent larger than before the system was implemented.

The marketing data warehouse also allows Western and Southern Life to determine:

  • When people are most likely to want Western and Southern’s products and services
  • Which events trigger customers’ interest so their needs can be anticipated
  • How customers prefer to do business
  • When and why customers are most likely to leave Western and Southern Life
  • Who are the most profitable customers and how to find more customers like them
  • Who are the most profitable agents and how to find more agents like them

The marketing data warehousing initiative has proven to be a key enabler in allowing Western and Southern Life to reach its vision.

Conclusion
Prior to Western and Southern Life’s data warehouse, marketing campaigns generally focused on saturating an undifferentiated client population with direct marketing materials so there would be enough responses to generate an acceptable return.Western and Southern Life’s strategy has changed from a mass-marketing approach to a focused, targeted approach. By using data mining with a data warehouse to identify the most likely responding segments, Western and Southern Life is able to achieve phenomenal ROI results, while only marketing to a limited portion of the customer base. This targeted approach dovetails with Western and Southern’s larger corporate CRM initiative of offering relevant products to the right people at the right time.

Future plans for the data warehouse include providing support for developing data marts throughout the Western & Southern Financial Group. This integration will allow for a complete view of the client relationship across the wide array of financial services offered by the group.

This data warehouse initiative has shown that having the right project management, and implementing best practices along with having senior management’s full support can bring profitable results.

REFERENCES
“2000 Data Warehousing and Data Mining in the Insurance Industry: Floods of Information, Fountains of Knowledge.” Conning & Company Insurance Research and Publications, 2000.

Chesbrough, Thomas. “Data Warehousing in the Insurance Industry: Eight Practical Lessons for Insurance Data Warehousing,” DM Review (October 1999), www.dmreview.com/master.cfm?NavID=55&EdID=1471.

Gaskin, Barbara C. “Realizing the Strategic Value of Data Warehouses,” Information Executive, Vol. 2, Iss. 4, (April 1998), 1-4.

Marks, Wendy and M. N. Frolick. “Building Customer Data warehouses for a Marketing and Service Environment: A Case Study,” Information Systems Management, Vol. 18, No. 3, (Summer 2001), 51-56.

Moncla, Brenda. “Data Warehousing: Managing Your Most Valued Asset,” DM Review (June 1998), 68-71.

Moss, Larissa and Sid Adelman, “Data Warehouse Goals and Objectives: Part 1,” Online posting, DM Review (September 1999), www.dmreview.com/master.cfm?NavID=55&EdID=1365.

Moss, Larissa and Sid Adelman. “Data Warehouse Goals and Objectives: Part 2: Short Term Objectives,” Online posting, DM Review, (October 1999), www.dmreview.com/master.cfm?NavID=55&EdID=1456.

Parzinger, Monica and Mark N. Frolick. “Creating Competitive Advantage Through Data Warehousing,” Information Strategy: The Executives Journal, Vol. 17, No. 4, (Summer 2001), 10-15.

Mark Kreyenhagen -

Mark Kreyenhagen is the Director of Database Marketing at Western-Southern Life. He is responsible for building, enhancing, and deriving business value from the data warehouse that he created.

Connie B. Robbins -

Connie B. Robbins is a Data Warehouse Analyst at Western-Southern Life. He is responsible for deriving business value from the data warehouse through data mining and marketing support.

Dr. Elaine Crable -

Dr. Elaine Crable is Chair of Information Systems at Xavier University. Her specialties include database management, data warehousing, business intelligence, and data mining.

Dr. Mark N. Frolick -

Dr. Mark N. Frolick is the Western & Southern Financial Chair of Information Systems at Xavier University. His specialties include data warehousing, executive information systems, e-business, cycle time reduction, systems analysis and design, and the diffusion of information technology in organizations.