On-Line Analytical Processing at Washtenaw Mortgage Company

On-Line Analytical Processing at Washtenaw Mortgage Company

John H. Heinrichs and William J. Doll

Idea Group Publishing

IDEA GROUP PUBLISHING IGP

1331 E. Chocolate Avenue, Hershey PA 17033-1117, USA Tel: 717/533-8845; Fax 717/533-8661; URL-http://www.idea-group.com

16##&'

On-Line Analytical Processing at Washtenaw Mortgage Company John H. Heinrichs Wayne State University, USA William J. Doll University of Toledo, USA

In an ever-changing, competitive marketplace, executive information systems (EIS) promise the ability to simultaneously assess factors in both the internal and external environment, enabling a timely competitive response. EIS are enjoying a renaissance due to the recent emergence of on-line analytical processing (OLAP) capabilities. OLAP’s power, flexibility and ease of use supports mental model (knowledge) creation better than traditional executive information systems. This case study allows you to examine the usefulness and ease of use of OLAP technology for strategic market analysis at Washtenaw Mortgage Company, a firm in the mortgage wholesale industry. The key to improving competitive performance is not the technology, but rather, how the technology is utilized to focus management’s analysis. Gaining strategic insights requires three ingredients – people, process, and technology. A three-stage process used for implementing an OLAP strategic market analysis application is presented. OLAP technology marks an evolutionary improvement in EIS software. The potential of this technology, however, is not likely to be realized without a better understanding of the process for achieving management focus.

BACKGROUND The mortgage wholesale industry was selected for this case study because organizations in this industry experience rapid market changes and are capable of developing products quickly to target specific market niches. To prosper in this industry, organizations must understand the requirements of the distribution channels as well as the ultimate consumers. Copyright © Idea Group Publishing. Copying without written permission of Idea Group Publishing is prohibited.

2 Heinrichs & Doll

In 1996, Wholesale Access, a mortgage research company, reported statistics highlighting organizations in this industry. The report identified organizations with growing and declining loan volume activity during an increasing interest rate environment (LaMalfa & Olson, 1996). A firm that had declining loan volume activity during this period was selected. The CFO of the firm was contacted and this case study was proposed. The requirements for the firm to participate in this case study included: (1) a willingness to provide an overview of the firm’s strategy, marketing direction, and performance metrics, (2) an agreement to provide access to the firm’s historical sales data, and (3) the ability to provide a cross-functional management team to explore OLAP and its ability to enhance strategic market analysis. The selected firm was Washtenaw Mortgage Company (WMC) located in Ann Arbor, Michigan. The volume of loan activity had declined and the firm’s objective was to return to a growth position. The contacted firm has an experienced senior management team and employs over 160 people in various departments (refer to Figure 3 in the Appendix). The organization conducts business in over 30 states through 1000 correspondents (independent agents) and through direct phone sales (retail). The correspondent network is managed by 20 account executives reporting to the vice president of sales & marketing. The firm offers a variety of products and services including 30-year mortgages, 15-year mortgages, adjustable rate mortgages, and balloon mortgages. Before beginning the case study, the CFO received a general briefing and demonstration of the OLAP tool, its analytical features and performance capabilities. The purpose of the briefing was to determine if the OLAP tool could be utilized in this organizational setting and to determine if the OLAP tool had potential value to the firm as perceived by the CFO. In this briefing, the CFO commented that OLAP had applicability, especially in the analysis of markets and products, and agreed to participate in this case study.

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u SETTING THE ST Ap GEP STA u o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop The future business climate will be characterized by increasing complexity and market diversity and by market and technological change requiring management to cope with market and product uncertainty (Gerwin, 1993; Doll & Vonderembse, 1991). To meet the competitive challenges of the next century, the organization’s knowledge workers will require improved tools for understanding changing market environments and customer requirements (Leidner & Elam, 1993; Peters & Brush, 1996). Historically, forecasting tools have been used to analyze the competitive environment and implement strategy with the objective of helping management predict the future of a given market segment or product line (Clark, 1992). These tools aided in reducing decision uncertainty by providing a degree of confidence to those decisions related to the success of market segments and product lines. Today, a key issue facing senior information system (IS) executives is the implementation and utilization of executive information systems (EIS) to aid in decision-making (Palvia, Rajagopalan, Kumar & Kumar, 1996). EIS are defined as computer-based information systems that integrate information from both internal and external data sources enabling managers to access, monitor, and request information of key importance to them via various presentation formats (Leidner & Elam, 1993; Millet & Mawhinney, 1992). Despite the importance senior IS executives place on these systems, only 32% of the organizations surveyed were satisfied with the information from their existing EIS (Li, 1995).

On-Line Analytical Processing 3

Expert practitioners have stated that EIS applications are high-risk/high-return projects and expensive to implement with almost a 25% failure rate (Kuehn & Fleck, 1991; Millet & Mawhinney, 1992; Rainer & Watson, 1995). Case studies and various monographs yield a variety of reasons for this situation, including issues of technical complexity, lack of management focus, inflexibility, and difficulty in assessing benefits (Belcher & Watson, 1993; Kuehn & Fleck, 1991; Millet & Mawhinney, 1992). The complexity and inflexibility issues include the requirement of structured and often inflexible data structures, predefined presentation views, and limited ad-hoc calculation capabilities. Technical complexity, inflexibility, and lack of management focus are possible explanations of the 25% project failure rate and the 32% satisfaction rate with the information provided by EIS applications that are implemented. A new type of EIS capability called On-line Analytical Processing (OLAP) is being made available to managers. These OLAP features are providing a major evolution in EIS capability. OLAP tools improve speed of information access and ad hoc calculation capability. OLAP supports multiple analytic dimensions simultaneously. These improvements in presentation, flexibility, and speed enable the market and product analysis to proceed at “the speed of questions”. Previous case studies on executive information systems focused on traditional (nonOLAP) applications in primarily large organizations (Belcher & Watson, 1993; Kuehn & Fleck, 1991; Rainer & Watson, 1995). While important in expanding our understanding of EIS capabilities and organizational benefits, these studies provide few insights as to whether OLAP technology is likely to reduce the EIS failure rate or enhance user satisfaction. Vandenbosch & Higgins (1995) call for case studies to identify the conditions that affect how effectively executive information systems are utilized. At present, little is known about whether, or under what conditions, OLAP capabilities (i.e., multiple analytical dimensions and moving at the “speed of questions”) can contribute to the effective utilization of executive information systems.

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P p u o r Competitive Advantage and Knowledge Creation G a e d I t h g i r y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop Nonaka (1991) tells us that the only certainty is uncertainty and that the one true source of competitive advantage is knowledge. Therefore, achieving superior performance is dependent upon the ability to make sense of uncertain environments. To develop that knowledge and to gain insight into uncertain market situations, executives seek to manage uncertainty by bringing evidence to bear. By facilitating the analysis of evidence, EIS fosters the creation and maintenance of the manager’s mental model (knowledge), resulting in competitive advantage. The enhanced model presented by Vandenbosch and Higgins (1995) links EIS characteristics, mental model (knowledge) creation and competitive performance (see Figure 1). EIS are technology-enabled tools to assist in making sense of the environment and to provide insight in an expeditious manner. EIS characteristics include analysis capability and usage characteristics. Analysis capability is a design characteristic (i.e., features of the software design); whereas, ease-of-use and information value are usage characteristics (i.e. characteristic dependent upon the quality of the EIS utilization). The OLAP design characteristics related to mental model (knowledge) creation are multi-dimensional analytical capabilities, presentation management, guided analysis, and dynamic creation of new analytical capabilities. The new EIS (OLAP) has enhanced capabilities in each of these four

4 Heinrichs & Doll

Figure 1: The OLAP Model

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e Analytical Capabilities d I Multi-Dimensional t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I Presentation Management t h g i r y p Co EIS Characteristics

Design

Usage

>

Mental Model Creation

Competitive

> Performance

>

• Multi-Dimensional Analystical • Ease of Use Capabilities • Presentation • Information Management Value • Guided Analysis • Dynamic Creation of New Analytic Capabilities

Managerial Insight

areas. Figure 1 depicts the dynamic relationship between design characteristics, usage characteristics, and the mental model creation required for competitive performance. Based on their analysis, managers can pose new questions and gain insights into their data by quickly changing dimensions and measures. OLAP enables this cycle of analysis and managerial insight to proceed at the “speed of questions.”

In the early 1980s, Dr. Codd developed the concepts that form the rules and standards for relational database technology. Relational database concepts are used in transactional processing systems and traditional EIS applications. In the early 1990s, Dr. Codd also defined the rules and features for the database technology to be used in OLAP (Codd, Codd & Salley, 1993; Newing, 1994; Radding, 1994). OLAP became the term synonymous with this database technology and is defined as the name given to the dynamic enterprise analysis required to create, manipulate, animate, and synthesize information from enterprise data models. This definition of OLAP database management capability includes the ability to: (1) discern new or unanticipated relationships between variables, (2) identify the parameters necessary to handle large quantities of data, (3) create an unlimited number of analytic dimensions, and (4) specify cross-dimensional arithmetic and business logic expressions. These capabilities are thought to address the perceived technical complexity/flexibility limitations inherent in traditional EIS software. Nigel Pendse (1997) summarized these OLAP rules and features into a simple fiveword expression labeled ‘Fast Analysis of Shared Multidimensional Information’. In this expression, fast implies that the majority of responses requested by the user are handled in less than five seconds. Analysis means that the system can handle statistical analysis and business logic relevant to the organization. Shared means that security requirements are addressed. Multidimensional means a multidimensional conceptual view of the data can be obtained. Information means all required data and calculations are available when required by the user.

The presentation capability can aid in mental model creation by linking problem representation with the decision-making task (Vessey, 1994). Either graphs/charts or tables can represent the problem. Graphs/charts are spatial problem representations that emphasize

On-Line Analytical Processing 5

the relationship in the data, whereas, tables emphasize discrete data values (Vessey, 1994). Depending upon the required decision-making task, usage of one or the other problem representation methods or both in combination is appropriate.

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u Dynamic Creation of New Analytic Capabilities P up o r G a e d I t h g i r y p o g C n i h s i l b u Using OLAP for Strategic Market P Analysis up o r G a e Id t h rig y p Co g n i h is CASE DESCRIPTION ubl P p u ro G a e d I t h g i r y Cop Guided Analysis

Evocation of relevant information is a critical aspect of decision-making (Browne, Curley, & Benson, 1997). OLAP’s guided analysis feature provides a starting point (a prompt) to create the manager’s mental model. The powerful OLAP database engine interacts with presentation and guided analysis features to create OLAP’s enhanced capabilities for evoking knowledge. With OLAP evoking knowledge, the manager strives to understand and gain additional insight into the information being presented. The information presented in response to the starting questions can spark additional questions from the manager, leading to additional insights.

At times, pursuing the implications of a new managerial insight might require users to make on-line changes in the software’s analytical capabilities without analyst intervention. OLAP enables managers to change analytical dimensions and/or utilize new cross-dimensional business logic or arithmetic/statistical expressions. Figure 1 illustrates how managerial insights from new mental models can lead to the dynamic creation of new analytical capabilities. The new analytical capabilities, in turn, enhance analytical, presentation, and guided analysis capabilities. This dynamic capability enables managers to “move at the speed of questions” rather than being delayed by the need for analyst intervention to change analytical dimensions.

OLAP is thought to enable dynamic enterprise analysis with a power, flexibility and ease of use that enhances the mental model creation of managers. Market analysis is especially well suited to the application of OLAP. Day (1994) and Zirger & Maidique (1990) describe the importance of clearly understanding markets and product offerings, the need for a “value to customer” focus, and the need for continuous learning about markets in a dynamic and uncertain environment. Transaction-based systems in most firms usually provide a wealth of detailed information for market analysis. OLAP’s powerful multidimensional database and presentation capabilities are well suited for coping with the size and complexity of the typical market analysis task.

The objective of this case study is to gain insight into the problems and opportunities related to the use of OLAP for market analysis in small or mid-size organizations. To explore this objective, the following questions were identified. 1. For strategic market analysis, can the analytic dimensions, guided analysis categories, and performance measures used in the OLAP application be easily identified and implemented? 2. Is the process of implementing a pilot OLAP market and product application difficult and time consuming?

6 Heinrichs & Doll

3. Is the guided analysis capability effective in evoking questions from managers? 4. Do managers perceive the OLAP application as an easy-to-use and useful (i.e., high information value) tool for market and product analysis? 5. Can managers effectively utilize OLAP’s capabilities to dynamically create new analytical dimensions/performance measures for market and product analysis?

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop To answer these questions, the firm, Washtenaw Mortgage Company, was contacted and this case study proposed. The CFO of WMC agreed to be the liaison and review the OLAP output at each stage for validity and accuracy. The authors of this case study agreed to provide the OLAP application tools, the installation expertise and the facilitation skills for the management meetings. This division of responsibilities enabled the potential perceived complexity issue to be controlled and minimized. The case study was conducted in three unique stages (see Figure 2). In the first stage, the analytical requirements of the firm were identified and confirmed. Identifying these requirements included three major tasks: setting study objectives and identifying key business areas to analyze, defining the analytic dimensions and performance measures the firm typically uses to assess market and product performance, and reviewing the capabilities of the firm’s information technology function. The second stage of the study focused on developing a pilot application system to be presented in a senior management meeting. Developing the pilot application system included three major tasks: development of the guided analysis application, loading the dimensional hierarchy and analytic data, and facilitating the senior management meeting to make preliminary assessments of perceived value and usability. The third stage was achieving management focus. This was accomplished by performing several tasks. The first task was to expand the available internal data for market and product analysis. Next, statistical analysis on dimensions and performance variables was performed utilizing SPSS statistical software. External data (demographic census data) was loaded into the system to enable comparisons and identify market niches. Finally, the trending and ‘detect and alert’ capabilities of the OLAP software were enabled. These tools enabled management to focus its analysis on significant and relevant issues. Each stage of this process required a greater time and resource commitment from both teams. The sponsoring senior executive reviewed progress at the completion of each stage before authorizing work to proceed on the next stage. The executive also verified the performance measures, business logic and formulas used in the model and made recommendations regarding how the data should be initially presented. Figure 2: Implementation Process

On-Line Analytical Processing 7

Stage 1: Analytical Requirements The first step in this stage was to set objectives and identify the key business area (function or process) to be analyzed. Management decided that the primary objectives for using OLAP were to analyze market data, identify opportunities, and help the firm return to a growth position. Sales and marketing had responsibility for developing multiple channels for product distribution, enhancing product offerings, expanding investor options, and analyzing market trends to ensure competitive offerings were available. Thus, attention was focused on the sales and marketing function. In the second step, the dimensions for analysis and performance measures were determined in a two-hour meeting with the CFO. During this meeting the perceived problems and opportunities facing the company were reviewed. The types of information used in answering these issues were identified from current management reports and on-line computer queries. A preliminary demonstration of the capabilities of the OLAP tool was completed as this facilitated the understanding of terminology and proposed usage of the captured data. (Given the access to an experienced financial officer, an analyst familiar with OLAP could expect to facilitate the identification of dimensions and performance measures in a single meeting.) The dimensions selected for analysis were geographical area, investor type, distribution channel, customer, product, time period and version (e.g. actual, budget, forecast, or benchmark). (Refer to the appropriate sections in the appendix for detailed information about each dimension.) Region and state classified the geographic area dimension. This classification matched the census bureau’s classification to facilitate analysis and comparison of data. (For more information, see the U.S. Census Bureau web site at http:// www.census.gov). The investor types dimension included the members Federal National Mortgage Association (for more information on Fannie Mae see http://www.fanniemae.com), Federal Home Loan Mortgage Corporation (for more information on Freddie Mac see http:/ /www.freddiemac.com), Independent National Mortgage Corporation (INMC), Banker’s Credit (BCD), and Pelican National Bank. The distribution channel dimension included both direct phone sales and the correspondent network. The various correspondents were associated with their account executives to facilitate analysis. The customer dimension included eleven rate lock categories that were chosen by the consumer. For example, a consumer could request a 15 days or a 45 days rate lock on their prospective loan. The rate lock category is important because the longer the number of days chosen, the greater the risk WMC assumes if interest rates increase. The products were summarized into seven major categories for the product dimension including 30-year mortgages, 15-year mortgages, adjustable rate mortgages, and balloon mortgages. (Management felt analysis by these summarized products would suffice for the initial pilot application.) For the time dimension, the time periods chosen included a rolling 12 months, current year, prior year, period-to-date (PTD), prior PTD, year-to-date (YTD) and prior YTD. Each of these groupings could be “drilled down” into the quarter, month, or week. During the pilot application, only the actual company data for the version dimension were loaded. The performance measures were categorized according to volume characteristics and profitability characteristics. The volume characteristic measures included amount borrowed and number of loans. The profitability characteristic measures included estimates of net cost, weighted average service fees (WASF), and service release premiums (SRP). Estimation was required to indicate the value of the current portfolio before the servicing rights were sold

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop

8 Heinrichs & Doll

to an investor group. In the third step, the skill of the IS staff, the adequacy of historical databases, and the hardware platforms were assessed. The IS staff had strong and broad technical skills. The IS function was led by the CFO who is a certified public accountant (CPA) and has an IS background as a computer auditor with Ernst & Young. The IS staff consisted of an IS manager, a program manager, six programmers, a personal computer support team, and system operators. WMC utilizes a state-of-the-art computer systems with over 500MB of main memory, 38GB of on-line storage, 190 direct connected devices, remote access for the account executives and network access for the correspondents via their WMCNet. The system operates a 24-hour a day, seven day a week schedule. It was concluded that the firm had the required hardware infrastructure (i.e. power, cooling, network) to handle the additional processing requirements of the OLAP application. It was determined that the staff would be able to provide the support necessary to extract the data and maintain the OLAP application on an ongoing basis. Also, it was concluded that the firm could provide the current and historical summary data required by OLAP. Upon completion of the analytical requirements stage, the results were reviewed with the CFO. The CFO felt that the necessary information was collected and the study should proceed.

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop Stage 2: Pilot Application For OLAP software, Comshare Sales Analysis (CSA) Version 1.3 running under Microsoft Windows Workstation NT Version 4.00 was chosen (for additional information on this decision support software product, see the Comshare Inc. web site at http:// www.comshare.com/Applications/cdolap.asp). The hardware platform utilized was an IBM compatible PC with a single 300MHz Intel Pentium processor, 64MB of RAM and 4.7GBdisk storage. CSA provides sample categories with guided analysis questions and the ability to customize category terminology. It also enables guided analysis questions to be added. These prompting questions, which are used in knowledge evocation, were added to each category to focus on the problems and issues identified in the analytical requirements stage. The prompting questions were designed to focus the manager’s thoughts and ensure a basic understanding of the analytic dimension and key performance indicators. The prompting questions start the analysis process but do not control the manager’s direction of questioning. They act as entry points to initially focus the manager’s thoughts in a particular area (refer to the appendix for Pareto analysis section for a summary). Examples of the prompting questions include: • Which are the fastest/slowest growing markets? • What is the product mix by market? • What is the product performance improvement by account executive? • How well are the currently offered products performing in the marketplace? • How well are the chosen channels marketing the products and service offerings? • In which markets are the various products selling and in what amount? • Can profit and sales trends be identified? Elements that were provided for the manager on each view were the future performance trend line, the mean line for a higher level aggregation, and simultaneous display of key performance measurements. Views for the manager included loan volume by state, product sold by state, product sold by time, product sold by channel, and product classifications.

On-Line Analytical Processing 9

The second step was to load the data into the pilot application. The defined multidimensional database required 407.7 megabytes of storage. The geographical area and investor type, distribution channel, customer, product, time period and version dimensions created 200, 1,100, 100, 150, 300, 5 members respectively. This represents a database of average complexity for market and product analysis in a firm of this size. The database was loaded and calculated, without incident, in approximately three hours. The third step of the pilot application was to facilitate a senior management meeting utilizing the pilot application system. Sample analysis graphic and tabular views for the application were created and reviewed with the CFO for completeness and accuracy. Following the CFO’s approval the meeting was held. Team members at the meeting were the Chairman of PN Holding, VP & CFO PN Holding, President WMC, VP Marketing WMC, Loan Pricing Manager WMC. The CFO focused the discussion of the senior managers on their normal business agenda. The difference in this meeting was that the pilot application provided answers to questions posed by various the managers. A sample question was “What is the impact to profitability if the commission paid on loans over $100,000 were increased?” This meeting structure ensured that tool usage did not interfere with questioning and potential insight into the data. The completion of Stage 2 confirmed the ability of the pilot application to provide answers to the questions posed by management. The multidimensional capability and the dynamic calculations enabled managers to identify and understand additional issues. This, coupled with the capability of automatic detection of out of variance conditions, encouraged the managers to ask additional questions. Based on this rich information source, management generated a number of options and business scenarios. The examination of these scenarios provided new insights into their business. Focusing on the answer to some of these questions, however, necessitated the acquisition of additional data.

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop Stage 3: Achieving Management Focus Achieving management focus was considered to be critical to the successful use of OLAP for executive decision-making. OLAP is an extremely powerful analytical tool that enables managers to explore a wide variety of issues. However, effective action often requires the whole management team to develop a shared understanding of the performance issues confronting the firm. Experience suggested that this rich OLAP capability to explore a variety of issues must be focused on key performance indicators. The techniques for achieving this focus were: expanding the database to include missing data needed for the analysis, conducting statistical tests, benchmarking with external data sources, and using Pareto analysis. The first step of this stage was the expansion of the data set to include multiyear market and product records. Because the multidimensional data definition used in the pilot was comprehensive and the data acquisition rules did not have to change, adding additional data and recalculating required only a few analyst man-hours. The computer time required to load the additional data was similar to the initial pilot load and the disk storage requirements did not expand appreciably. It was estimated that, on an ongoing basis, monthly updates of the database would take about three hours on this hardware platform. This expansion facilitated further analysis of whether key questions and concepts uncovered in the management review meeting could be consistently answered. The second step was to conduct statistical tests using SPSS 7.5 for Windows’ routines

10 Heinrichs & Doll

for non-parametric correlation, cross-tabulation, and multivariate analysis of variance. (For additional information, refer to SPSS Inc. web site at http://www.spss.com.) To determine whether distinctions within dimensions (i.e., among categories) were significant, descriptive and inferential statistics were reported at a macro level (i.e., a level of aggregation appropriate for senior management). Where the differences were not significant and substantial, the analysis was simplified to focus on the key issues. For example, the results indicated that loans over $100,000 were substantially more profitable than loans of a smaller dollar amount. This focused management attention on whether larger commissions should be paid on these loans to increase production. Comshare Sales Analysis software provides statistical capabilities, but CSA does not include cross-tabulation and multivariate tests. Because an adequate statistical analysis was so important in achieving management focus, using more powerful statistical analysis tools to facilitate inferences was recommended. Without careful statistical analysis, considerable management time can be spent arguing differences that are not statistically significant (i.e., due to chance). This interferes with achieving management focus. The third step was to locate and enter external data for benchmark comparisons. External housing data and homeowner demographics were obtained from the census bureau (for more information, see the U.S. Census Bureau Web site at http://www.census.gov/hhes/ www/housing.html). This external data was matched to similar internal statistics from WMC to facilitate comparison. This provided a validation of the management strategy and focused the analysis. For example, the demographic information would enable management to determine how WMC’s percentage of loans over $100,000 compared to the market across geographic areas. The fourth step in achieving management focus was the use of Pareto analysis utilizing the 80-20 rule. Pareto analysis means focusing on the most important information first. The rule that 20 percent of your activities contribute 80% of your performance was used to focus attention on the most important activities. The defined analytical dimensions were analyzed by the volume and profitability performance measures. Members (categories) that comprised 80% of the defined measures were included in the analysis. Pareto charts were developed for management review. These charts helped to focus management attention. For example, management found that they should, perhaps, focus on the seven states that contributed 80% of the loan volume activity and compare these states’ success to the expansion efforts in other states. The four steps in this stage helped develop a shared mental model among the management team. The guided analysis capability was effective in evoking questions from managers. Expanding the database enabled the management team to answer key questions and think of new questions. It is believed that OLAP’s dynamic capability enabled managers to remain in mental model creation for a longer period of time than would have occurred with a traditional executive information system. The statistical test focused attention on significant distinctions and helped gain management consensus on which distinctions were truly important. Benchmarking enabled market comparisons and Pareto analysis focused attention on the critical areas for action. These techniques were considered helpful in focusing management attention and gaining consensus.

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop

On-Line Analytical Processing 11

CURRENT CHALLENGES F ACING THE FA ORGANIZA TION ORGANIZATION

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e Development d I Application t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I Strategic Insights t h g i r y Cop In today’s competitive global marketplace, the effective utilization of executive information systems for market and product analysis can be essential in making sense of uncertain and rapidly changing market environments. Such applications can help organizations identify market opportunities and threats, evaluate their strengths and weaknesses, and formulate appropriate action plans. A better understanding of the reasons for project failure or user dissatisfaction can help organizations reduce the risks associated with these high payoff strategic applications. This case study focused on Washtenaw Mortage Company’s experience in utilizing OLAP technology (i.e., Comshare Sales Analysis software) for strategic market analysis. One should be cautious in attempting to generalize this experience to other firms, applications, or software vendors, especially vendors that do not provide the “top of the line” capabilities inherent in the Comshare software. Nevertheless, this case study was valuable in gaining firsthand insights into whether technical complexity, inflexibility, and lack of management focus are likely to be important factors explaining the success or failure of OLAP applications.

CSA is a rich, flexible, and powerful application tool with high functionality and low perceived complexity. The application was relatively simple to install, customize, and operate. CSA software represents a substantial improvement in ease of installation, design, and use. The high EIS failure rate reported in the literature may have been caused by the technical complexity of traditional EIS applications. However, the applications based on Comshare’s software are unlikely to fail because of technical complexity. The Comshare’s design characteristics (see Figure 1) were an improvement in flexibility over traditional EIS software. The analytical categories, guided analysis questions, and performance measures used in this OLAP market and product analysis application could be easily identified and implemented. The management team tacitly knew the analytical categories required and, upon questioning, were able to provide an explicit list of the key performance measures. The guided analysis capability was effective in evoking questions from the managers. It acted as a starting point and a review of underlying assumptions for the management team. Management used the dynamic calculation capabilities to create ratios and used the multidimensional capability to summarize information at new levels, thereby gaining additional insights. The speed, flexibility, and ease of use of the software greatly enhanced information value, facilitating the process of mental model (knowledge) creation. The management team perceived CSA as easy to use and useful in decision-making. Executive information systems based on “top of the line” OLAP software such as CSA are unlikely to fail due to inflexible software or management’s concerns regarding its ease of use.

While the dynamic capabilities, guided analysis questions, and performance measures improved mental model creation, most of the insights generated had limited implications. The improved technical capabilities of OLAP did not, by themselves, evoke the kinds of

12 Heinrichs & Doll

managerial insights that would improve competitive performance. Improved management focus was essential to generating strategic insights. Strategic insights have many causes—people, process, and technology. They were, perhaps, most dependent upon the leadership of the chief financial officer and the management team. Many of the strategic insights emerged from the process (see Figure 2). The statistical analysis, benchmarking, and Pareto analysis activities in Stage 3 helped management question existing beliefs and focus the analysis on important issues. Most of the strategic insights emerged during these later activities. A broad range of skills is required to use these analytical techniques appropriately and to communicate the findings effectively to the management team. Many small and medium size firms may not have in-house staff with the skills to facilitate this process. Rather then buying the technology and using an in-house facilitator/analyst, smaller firms may find that an experienced consultant can be used on a periodic basis. Without more attention to the process and the required skill set, many organizations may be purchasing strategic information technology that they can not effectively utilize. Challenging existing assumption is not done by technology alone. Managers, working with an analyst (facilitator) with appropriate skills, need a process of inquiry that helps them to achieve the focus required for strategic insights. Using the technology and following the process identified in Figure 2 should help management understand the data, so that they can develop insights to improve their performance. EIS iare enjoying a renaissance because of the emergence of OLAP technology. This technology can effectively address concerns that technical complexity and inflexibility are responsible for EIS project failure or dissatisfied users. However, the technology does not, by itself, result in the kinds of strategic insights that improve competitive performance. Strategic insights require improved management focus. Achieving this focus depends upon the people and the process of inquiry, not just the technology. The potential of this technology is not likely to be realized without an increased emphasis on the process of achieving management focus.

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h APPENDIX g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop This appendix contains information that may be useful in further understanding Washtenaw Mortgage Company. The sections are: • Business Analysis Review • Pareto Analysis Summary • Analysis Dimensions • Product • Time • Channel • Customer • Geography Each of these sections provides information to highlight the various analysis capabilities.

On-Line Analytical Processing 13

Appendix 1

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop Figure 3: Organization Chart

Figure 4: Value Chain

14 Heinrichs & Doll

Appendix 2. Organizational Structure

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop Business Analysis Review

This section highlights the percentage increase in the volume of loan production. Similar analysis could be performed on all performance metrics. It is based upon calculations from the 1st quarter as compared to the 4th quarter. Account Executive Analysis • AE 22 grew at a 68% rate • AE 25 grew at a 46% rate • AE 21 grew at a 36% rate

Market Analysis • Illinois grew at a 79% rate • Michigan grew at a 66% rate • Indiana grew at a 41% rate • Florida fell at a 5% rate • Georgia fell at a 25% rate • Louisiana fell at a 81% rate

Product Analysis • the 15-year mortgage grew at a 31% rate • the 30-year mortgage grew at a 19% rate • the average growth rate for all products during the year was 27% Product Mix Analysis • for 15-year loans the highest growth rate by state was • Indiana grew at a 31% rate • North Carolina grew at a 30% rate • Georgia grew at a 28% rate • • • •

for 15-year loans the lowest growth rate by state was Wisconsin grew at a 11% rate Illinois grew at a 11% rate Michigan grew at a 11% rate

• • • •

for 30-year loans the highest growth rate by state was Alabama grew at a 77% rate Florida grew at a 77% rate Louisiana grew at a 74% rate

• • • •

for 30 year loans the lowest growth rate by state was Wisconsin grew at a 59% rate Indiana grew at a 55% rate Illinois grew at a 55% rate

On-Line Analytical Processing 15

Rate Lock Analysis • The 15 day rate lock showed a 38% increase

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop Appendix 3. Pareto Analysis:

This section highlights the members of the category that contribute over 80% of the volume of loans for the calendar year. Account Executive • 6 Account Executives – 4, 25, 21, 19, 0, 27 • 273 Correspondents Market Analysis • 7 States – MI, OH, IN, FL, GA, IL, KY Product Analysis • 30 year & 15 year mortgages Rate Lock Analysis • 5 days – 40% • 30 days – 35%

Appendix 4: Product Dimension

16 Heinrichs & Doll

Appendix 5: Time Dimension

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop

On-Line Analytical Processing 17

Amount Borrowed

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop Total Amount

Number

Between

of Loans $75k—$125k

Less Than

Quarter

Borrowed

Mean

1 Q 1997

$140,712,995

1620

36%

$86,860

70%

81%

2 Q 1997

$138,433,986

1630

36%

$84,929

72%

83%

3 Q 1997

$193,995,371

2092

36%

$92,732

63%

75%

4 Q 1997

$253,803,713

2555

37%

$99,336

57%

70%

Grand Total

$726,946,066

7897

37%

$92,053

64%

76%

Table 4: Amount Borrowed by Quarter

Appendix 6. Customer dimension - Rate Lock Analysis

Mean

Rate Lock (days)

Number of

Loan

Loans

Percent

Amount

40

0.5%

$93,697

1

1

0.0%

$66,956

5

3,318

42.0%

$89,638

10

1

0.0%

$164,600

15

1,454

18.4%

$90,356

30

2,378

30.1%

$88,852

45

292

3.7%

$130,893

50

239

3.0%

$93,043

60

166

2.1%

$130,614

75

2

0.0%

$83,550

90

2

0.0%

$99,450

120

1

0.0%

$74,500

180

2

0.0%

$85,650

0

550

1

0.0%

$128,2509

Total

7,897

100.0%

$92,053

Table 6: Rate Lock Analysis

$100,000

$120,000

18 Heinrichs & Doll

Appendix 7: Distribution Channel Dimension - Investor Type Analysis

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop Mean

Number of

Investor

Loan

Loans

Percent

Amount

Fannie

7,276

92.1%

$92,466

Freddie

113

1.4%

$88,477

INMC

306

3.9%

$89,546

BCD

21

0.3%

$73,976

PN Bank

181

2.3%

$84,055

Total

7,897

100.0%

$92,053

Table 7: Investor Type Analysis

Appendix 8: Distribution Channel Dimension-Account Executive Analysis Mean

Account

Number of

Executive

Loans

Percent

Loan

Amount

0

711

9.0%

$120,941

2

3

.0%

$106,935

4

1,583

20.0%

$82,490

10

5

.1%

$122,854

19

799

10.1%

$87,256

21

1,155

14.6%

$73,839

22

344

4.4%

$118,439

23

37

.5%

$124,477

24

458

5.8%

$86,874

25

1,383

17.5%

$103,850

26

214

2.7%

$78,443

27

599

7.6%

$84,750

28

97

1.2%

$79,043

29

5

.1%

$56,860

50

215

2.7%

$96,748

82

169

2.1%

$89,923

204

1

.0%

$39,000

221

2

.0%

$54,534

225

73

.9%

$127,891

227

33

.4%

$110,351

250

11

.1%

$98,456

Total

7,897

100.0%

$92,053

Table 8: Account Executive Analysis

On-Line Analytical Processing 19

Appendix 9: Geographic Dimension-State Analysis

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop State

Number of Loans

Percent

Mean Loan Amount

AL AR AZ CO DC FL GA IA ID IL IN KS KY LA MD MI MN MO MS MT NC ND NJ NM NV OH OK OR SC SD TN TX UT VA WA WI WV WY Total

162 1 2 88 11 668 536 80 2 274 674 32 227 152 42 2,412 87 66 13 86 212 2 1 23 6 1,552 31 12 134 5 66 23 9 55 2 107 28 14 7,897

2.1% .0% .0% 1.1% .1% 8.5% 6.8% 1.0% .0% 3.5% 8.5% .4% 2.9% 1.9% .5% 30.5% 1.1% .8% .2% 1.1% 2.7% .0% .0% .3% .1% 19.7% .4% .2% 1.7% .1% .8% .3% .1% .7% .0% 1.4% .4% .2% 100.0%

$78,616 $56,900 $63,934 $117,263 $85,113 $85,973 $87,695 $72,360 $52,883 $123,498 $73,254 $75,437 $92,033 $80,647 $132,083 $106,656 $92,199 $85,629 $56,163 $70,791 $82,204 $45,606 $180,000 $74,024 $94,156 $81,526 $65,071 $80,746 $86,501 $69,152 $73,450 $73,489 $86,796 $127,493 $73,168 $102,132 $89,157 $75,209 $92,053

Table 9: State Analysis

20 Heinrichs & Doll

REFERENCES

g n i h lis b u P p u ro G a e d I t h g g i n r i y h s i l b Cop u P up o r G a e d I t h g i r y p o C g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h lis b u P p u ro G a e d I t h g i r y Cop Belcher, Lloyd W. & Watson, Hugh J. (1993). Assessing the Value of Conoco’s EIS, MIS Quarterly, 17(3), 239-253. Browne, Glenn J., Curley, Shawn P., & Benson, P. George (1997). Evoking Information in Probability Assessment: Knowledge Maps and Reasoning-Based Directed Questions, Management Science, 43(1), 1-14. Clark, Delwyn N. (1992). A Literature Analysis of the Use of Management Science Tools in Strategic Planning, Journal of Operations Research Society, 43(9), 859-870. Codd, E. F., Codd, S. B. & Salley, C. T. (1993). Providing OLAP to User-Analyst: An IT Mandate, Research Paper. Day, George (1994). The Capabilities of Market-Driven Organizations, Journal of Marketing, 58, 37-52. Doll, William J. & Vonderembse, Mark A. (1991). The Evolution of Manufacturing Systems: Towards the Post-Industrial Enterprise, Omega, 19(5), 401-411. Gerwin, Donald (1993). Manufacturing Flexibility: A Strategic Perspective, Management Science, 39(4), 395-410. Kuehn, Ralph & Fleck Jr., (1991). Robert A. Implementing and EIS in a Large Insurance Corporation, Journal of Systems Management, January, 6-17. LaMalfa, Tom & Olson, David (1996). Who’s Who in Wholesale 1995, Mortgage Banking, January, 48-56. Leidner, Dorothy E. & Elam, Joyce J. (1993). Executive Information Systems: Their Impact on Executive Decision Making, Journal of Management Information Systems, 10(3), Winter, 139-155. Li, Eldon Y. (1995). Marketing Information Systems in the Top U.S. Companies: A Longitudinal Analysis, Information & Management, 28, 13-31. Lim, Jeen-Su, Heinrichs, John, and Hudspeth, Lonnie (1999). Strategic Marketing Analysis: Business Intelligence Tools for Knowledge Based Actions, Pearson Custom Publishing, Needham Heights, MA. Millet, Ido & Mawhinney, Charles H. (1992). Executive Information Systems: A Critical Perspective, Information and Management, 23, 83-92. Newing, Rod (1994). Relational Databases are not Suitable for Management Information Systems: and that’s Official!, Management Accounting, Sept., 26-28. Nonaka, Ikujiro (1991). The Knowledge-Creating Company, Harvard Business Review, November-December, 96-104. Palvia, Prashant C., Rajagopalan, Balaji, Kumar, Anil, & Kumar, Ned (1996). Key Information Systems Issues: An Analysis of MIS Publications, Information Processing & Management, 32(3), 345-355. Pendse, Nigel (1997). The OLAP Report, London: Business Intelligence Ltd. Peters, Michael P. & Brush, Candida G. (1996). Marketing Information Scanning Activities and Growth in New Ventures: A Comparison of Service and Manufacturing Businesses, Journal of Business Research, 36, 81-89. Radding, Alan (1994). Is OLAP the Answer?, Computerworld, 28(51), 72-73. Rainer Jr., R. Kelly & Watson, Hugh J. (1995). The Keys to Executive Information System Success, Journal of Management Information Systems, 12(2), 83-98. Vandenbosch, Betty & Higgins, Christopher A. (1995). Executive Support Systems and

On-Line Analytical Processing 21

Learning: A Model and Empirical Test, Journal of Management Information Systems, 12(2), 99-130. Vessey, Iris (1994). The Effect of Information Presentation on Decision Making: A CostBenefit Analysis, Information & Management, 27, 103-119. Zirger, Billie Jo & Maidique, Modesto A. (1990). A Model of New Product Development: An Empirical Test, Management Science, 36(7), 867-883.

g n i h s i l b u P p u o r G a e d I t h g i r y Cop g n i h s i l b u P up o r G a e d I t h g i r y p o g C n i h s i l b u P up o r G a e Id t h rig y p Co g n i h lis b u P p u ro G a e d I t h g i r y Cop John H. Heinrichs is on the faculty in the Information Systems and Manufacturing Department at Wayne State University. He is completing his Ph.D. in Manufacturing Management at The University of Toledo. Prior to admission in the Ph.D. program, John spent twenty-two years in the Information Processing industry in roles of consulting, management, training, marketing, and technical support. He led the reengineering effort to provide a leading edge strategy development process for a Fortune 500 client consisting of organizational changes for flexibility, technology enhancements for speed, and structured process for discipline. He has co-authored a book on business intelligence tool usage for strategic marketing analysis and was recognized as a finalist in the Decision Science Institute 1999 Most Innovative Course of the Year award. His development of Quest, a customer satisfaction management reporting and analysis tool earned him recognition as well as an appearance in the 1994 SPSS annual report. He has presented at numerous conferences.

William J. Doll is a Professor of MIS and Strategic Management at the University of Toledo. Dr. Doll holds a doctoral degree in Business Administration from Kent State University. He has published extensively on information system and manufacturing issues in academic and professional journals including Management Science, Communications of the ACM, MIS Quarterly, Academy of Management Journal, Information Systems Research, Journal of Operations Management, Decision Sciences, Omega, Information & Management, Datamation, and Datapro. Dr. Doll has published on a variety of topics including computer integrated manufacturing, executive steering committees, top management involvement in MIS development, strategic information systems, information systems downsizing, and enduser computing. Dr. Doll has developed research instruments to measure end-user computing satisfaction and end-user participation in systems development.