Today’s competitive business environment requires companies to collect, analyze, and interpret enormous quantities of data to enable better, more informed decision making. Fortunately, there is an abundance of data available for most companies. However, without proper data quality and governance guidelines, these data might prove useless. Data quality is a necessity when extracting meaningful information. For, without it, a company may be left guessing at a direction or making skewed decisions based on flawed data. So, in order to achieve high quality, the data need to be accurate, available, understandable, and relevant to the problem being solved. Ultimately, impactful, data-driven decisions are necessary for organizational survival. Unfortunately, the decision making capabilities of a company are made significantly more complicated when there is a lack of data confidence. This lack of confidence adversely affects organizational performance. (IBM, 2008). Among all industries, the insurance industry may be the most reliant on quality data. Insurance companies derive their profit by predicting the cost of future losses their insureds Address correspondence to Thilini Ariyachandra, Williams College of Business, Xavier University, 3800 Victory Parkway, Cincinnati, OH 45207, USA. E-mail: will incur and creating pricing models to cover these and operating costs. Many insurance and financial service organizations utilize a process for creating predictions which consists of two primary steps: data acquisition and analytical processing. While not unique, it allows for data consistency and integrity. Unfortunately, not all processes are successful or timely. In this article, New Tech Insurance is introduced, an insurance company offering a vast array of insurance products. Like other insurers, New Tech Insurance employed a silo approach to data acquisition where it gathered data from several distinct systems. These different data sources often contained incomplete segments of the same data, as well as slightly different versions of the same data. Unfortunately, New Tech Insurance was tasked with processing and making important decisions with this poor quality data. In addition, New Tech Insurance struggled with data sourcing problems. New Tech Insurance spent more time assimilating data than analyzing due to a largely manual data integration process. A main cause was the manual task of correcting data quality problems. At the same time, the company’s research and development area was hindered by the lack of proper analytic tools, so advanced analytics were outsourced. In short, a consistent enterprise-wide data collection and analysis approach was lacking. In order to better capitalize on the wealth of information being collected, as well as to correct its current data collection issues, New Tech Insurance decided to create and implement a business intelligence competency center (BICC).This article provides an overview of New Tech Insurance, discusses the reasons and plans for initiating its BICC, and examines the success of the BICC.
NEW TECH INSURANCE AND BICC
New Tech Insurance is an insurance company operating in 23 states and offering property, casualty, and automotive insurances, as well as excess and surplus lines of insurance to its individual and business customers. The company was founded with the idea to make insurance a one-stop shop for customers and to operate through independent local agents. These 229 230 K. FOSTER ET AL. local agents are commissioned based and not employed by the company. Given their extensive data collection and analysis problems, New Tech Insurance saw the need to assess their data governance policies related to data collection and analytics. This led to the creation of a BICC. A BICC is a cross-disciplinary team assigned within a company who is charged with championing “BI technologies and standards, as well as the business alignment, project prioritization, management and skills issues associated with significant BI projects.” (Strange & Hostmann, 2003).
The success or failure of such a group is primarily dependent upon the composition and skills of the team. The BICC Initiative Prior to the creation of its BICC, New Tech Insurance had a single IT organization known as the IT Data Management Group. This unit managed all matters related to technology including developing reports for business users. Given that the IT Data Management Group managed all technology issues for the company, its workload steadily increased as the company grew. In addition, a constraint developed in the report creation process for business users. The problem centered on the IT Data Management Group not being able to create exactly what the business users truly needed in a timely manner. As such, there appeared to be a communication gap between the IT Data Management Group and the business users. This problem led to the IT Data Management Group creating reports for business users which were not used because the amount of time it took to create the reports exceeded their usefulness. So, instead of using the generated reports, business users sought the skills of experts within the organization to help generate reports which led to a lack of confidence in the IT Data Management Group and its ability to deliver quality and timely data.
New Tech Insurance faced a difficult dilemma as it had several key problems to address. They included:
• The ever growing work for the IT Data Management Group;
• The lack of effective communication between the IT Data Management Group and business users;
• The lack of data quality standards;
• The need for better analytic tools.
So, in 2008 a multi-departmental committee was formed to combat these issues. This committee decided to dedicate a team to bridge the gap between business users and the IT Data Management Group while simultaneously improving the analytical capabilities of the company. With topdown support from the Chief Financial Officer (CFO), the BICC was created.
For New Tech Insurance, they see their BICC as a hybrid organization combining technical and functional resources to manage information delivery. At the same time, the BICC collaborates within the business to improve analytical capabilities and deliver information solutions in support of its strategic objectives. The center was to serve as a business-led, internal consulting, analysis, and support organization. Individuals with the right skill sets were essential to take on a project of this size and importance.
For the BICC to be successful, the team required members to have the ability to communicate directly with business users as to gain a more holistic understanding of their needs, both in terms of content and time. In addition, these individuals required a technological background to understand how to accomplish the specific tasks required by the business users. Over 120 individuals within the organization were interviewed for the BICC team. What started as a team of one grew steadily as the responsibilities of the BICC expanded. The underlying mission of the BICC was to empower strategic decision makers to accomplish important business objectives through a structure with strict data governance and information delivery.
The BICC was to act as a liaison between analysts in the lines of business and the IT Data Management group. In order for it to work, the BICC needed to be business centric, rather than IT centric. This allowed it to better meet the needs of the lines of business it supports. Even though it was to be business oriented, the BICC still had technical responsibilities. Since there is overlap between the objectives of the IT Data Management and the BICC, there were varying opinions regarding roles and responsibilities. So, New Tech Insurance created guidelines regarding the responsibilities of the newly created BICC and clarified the responsibilities already in place for the IT Data Management Group (see Table 1). For example, the BICC was charged with developing the Cognos reports and dashboards, while the IT Data Management Group was charged with administering the created Cognos reports and dashboards. This separation of responsibilities was essential for two reasons. First, each group was able to follow its own mission and vision. Second, the separation of powers easily pacified conflicts when discrepancies surface. The BICC was aligned under the CFO in order to keep its focus on improving the company’s competitive business edge.
The IT Data Management Group remained a traditional IT team under the direction of the Chief Technology Officer. The CFO saw the critical success factors for the BICC to be its ability to effectively and efficiently manage its responsibilities and deliverables, while reducing the cost of providing information to the core analytic processes. Addressing Key Issues of Analytical Processing and Data Governance The BICC helped to address several key challenges that New Tech Insurance previously faced with its BI efforts under the IT Data Management Group. BICC 231 TABLE 1 BICC/IT Responsibilities BICC IT Data Management Semantic layer design Semantic layer design and build Cognos report and dashboard development Administration (SAS, Cognos, Infosphere) Data profiling Data profiling Data dictionary/Functional metadata Technical metadata Business reporting requirements ETL (IDL, EDW, DQ, etc.) Prioritizing change requests and development of (Cognos, SAS, IDL, ETL, and EDW work) Administrate EDW, IDL, Data marts Cognos/SAS end user training Cubes/OLAP development Sandbox starter set design Starter sets creation Data quality requirements for ETL Report scheduling Data governance and quality process Sandbox environment support Conceptual data model validation and design Conceptual/Logical/Physical data model creation, validation, and design Prior to the BICC, the IT group lacked timeliness with the delivery of data requested by business users. As a result, business users would turn to select power users within their own groups to assist in report creation.
At first these power users enjoyed the recognition of their sought after skills, but, in time, it became too difficult to accomplish their own responsibilities with this additional burden. The BICC recognized that providing business users with data they requested in a timely manner and helping them with their specific reporting needs increased their satisfaction significantly. The BICC also realized an opportunity to give the business a self-service data reporting application when they implemented Cognos, a business intelligence and performance management software suite. This software would ultimately allow business users to satisfy their own reporting needs as it provides novice users with the ability to extract, report, visualize, and analyze data.
This necessary acquisition helped to reduce business user dependency on the IT Data Management Group and focused their attention to the BICC for their information reporting needs. Prior to the creation of the BICC, no enterprise wide data governance policy existed. So, data analysts and business users experienced less than satisfactory results when analyzing data. Often, the development of reports and information processing were delayed by the lack of consistent, well-defined data. The BICC put forth a strong governance process to aid in the creation of impactful enterprise decisions. As such, the BICC sets-up and maintains definitions for data collected from the company’s business insurance operations. It schedules and facilitates periodic meetings for its oversight committee, the Enterprise Data Governance Council.
The Council ensures data stewards across multiple organizations are kept abreast of new and on-going changes to corporate data. A Data Governance Manager, who works in the BICC, was charged with chairing a data stewardship committee which is responsible for:
• Defining data across the enterprise;
• Improving data quality;
• Resolving data integration issues;
• Providing data usage policy and quality standards;
• Determining data security;
• Maintaining business rules applied to data and data retention criteria.
In addition, a data glossary was developed and deployed to help increase quality. This directory helped to centralize the definition of individual data elements and the business areas they support. New Tech Insurance also fell short with enterprise-wide quality standards and assessments. This resulted in root cause analysis and resolution being performed inconsistently and largely unchecked. The BICC put forth a goal to have the data quality processes be both proactive and reactive.
Proactive processes help prevent data quality issues from occurring as they enable business process improvement and early involvement in new information systems initiatives. Reactive processes, such as the collection and resolving of data quality issues, feed back into the proactive process by means of lessons learned. Now, the BICC provides support to data creation areas based on the knowledge gained through data quality issue resolution. It also prepares and publishes a data quality scorecard which enables the organization to gauge its progress on improving data quality. The BICC also works with the data stewards to help maintain data validation rules which enforce enterprise data standards, achieve key quality goals, and ensure root causes are addressed. Finally, the BICC addressed analytical process sharing. Prior to the BICC, analysis performed throughout the organization 232 K. FOSTER ET AL. was segregated by departmental units. By creating analytical silos, the company increased the amount of conflicting measures, redundant efforts, and decision making based on inaccurate information. The BICC catalogs the analytics dispersed throughout the company, validates the accuracy of the methods and results, and posts the analyses to an online repository which is available to all appropriate employees broadly. The BICC is also able to provide formatted data sets for statistical analysis. This enables actuaries to easily model data and design insurance lines with more precision. At the same time, self-service reporting through Cognos has allowed underwriters to decrease the amount of time needed to accomplish their tasks from weeks to minutes. These accomplishments have saved the company both time and money.
BICC BEST PRACTICES
The implementation of the BICC at New Tech Insurance addressed several key data and process issues in data governance and practices. These translated into major tactical and financial gains for New Tech Insurance. The success of the BICC posits several best practices for organizations attempting to implement BICCs. The first key is securing top down support from an executive sponsor willing to champion the project. This executive sponsor must be directly engaged with the project. This will help eliminate barriers and keep the project on point during difficult times. For, without an executive sponsor, a project’s chance of failure is much higher than with one. The second key is assigning tasks to the BICC away from other units to avoid redundancy and ease process bottlenecking. For New Tech Insurance, the transfer of responsibilities from the IT Data Management Group to the BICC helped to reduce data and task redundancy while allowing the BICC to take control of some responsibilities formally managed by the IT Data Management Group. The result was the BICC acting as the liaison between business users and traditional IT support. But, with the BICC acquiring more responsibilities, another group must lose some. This shift of responsibilities may have a “stepping on toes” reaction by those groups losing some of their responsibilities. However, in order for the BICC to achieve success, the shift of responsibilities is absolutely necessary. Those losing responsibilities will gain capacity and acquire other responsibilities which can be more focused to their core competencies. Third, organizations should seek and celebrate early wins. For a project of this magnitude, early wins allow for a gain in confidence in the BICC and its team. In addition, early wins also help improve the confidence of individuals within the BICC. “The most important implication of the progress principle is this: By supporting people and their daily progress in meaningful work, managers improve not only the inner work lives of their employees but also the organization’s long-term performance” (Amabile & Kramer, 2011, p. 12). Fourth, the organization must live the adage “teach them how to fish.” Strive to empower business users with self-service applications. A self-service approach allows business users to work more efficiently by completing tasks on their own and lifting the constraint on others who are often left to support when assistance is needed. Through self-service, employees gain efficiency and essential skills while the company saves money. Fifth, it is important to define architecture for different components of the BI infrastructure while continuously updating and maintaining data standards, methodologies, definitions, processes, tools, and technologies required to support BI. Without companywide standards, it is a challenge for different aspects of the company to work together. Sixth, a company needs to create a success roadmap with metrics which measure both the implementation and ongoing success of BI. It is also important to communicate successes and demonstrate how the BICC is enabling the organization to meet its BI goals. A common scorecard should be created to help visualize the progress made and the goals still unattained. Finally, for a proper and successful implementation of a BICC, soft skills prove to be more important than technical skills. For, as Eckerson (2005) states, “Successful technical teams score especially well on the ‘soft issues,’ such as the ability to communicate technical issues clearly, respond to business requirements, and develop desired functionality.” The individuals on the BICC team must possess the ability to communicate and understand business users’ needs and translate those needs into technical requirements.
Reflecting on the successes New Tech Insurance, it is clear that progress was due to the implementation of the BICC. However, as with any organization, no goal or implementation is ever complete. As technology advances, the BICC will improve upon its accomplishments. As Zeid (2006, p. 20) states, “It is a journey, not a destination. Changing internal culture and knowledge mechanisms is a slow process. You are embarking on an ongoing process that will pay significant dividends in the long term.” As it evolves, the BICC would like to integrate its services companywide and steer away from a pool of select power users.
In order to accomplish this goal, the BICC is developing a userfriendly dashboard system to replaces its current reporting tool. Dashboards will allow individual, novice users to manipulate data and solve problems in real-time, thus alleviating the load on the power-users. Ultimately, this should free up the IT Data Management Group to work on more important projects, and give regular business users more freedom in their work.
Kyle Foster is a Data Solution Consultant at 84.51 (formerly dunnhumbyUSA). He is responsible for the implementation of data strategy and relationships clients. Prior to joining dunnhumbyUSA, he worked as a consultant for Sogeti BICC 233 USA. Mr. Foster earned a Master of business administration with a concentration in business intelligence from Xavier University. Gregory Smith is an associate professor of MIS in the Williams College of Business at Xavier University. His primary research areas are the application of artificial intelligence in predictive analytics and data mining. He has published in the International Journal of Business Intelligence Research, IJPR, and Advances in Business and Management Forecasting.
Thilini Ariyachandra is an associate professor of MIS at Williams College of Business at Xavier University. Her main research area is BI and data warehousing. She has published in Decision Support Systems, Information Systems Management, Business Intelligence Journal, and the DM Review. Mark N. Frolick is a professor of MIS in the Williams College of Business at Xavier University and the holder of the Western and Southern Chair in Management Information Systems. He is considered to be a leading authority on business intelligence and has authored over 150 articles. His research has appeared in such prestigious journals as MIS Quarterly, Decision Sciences, Journal of Management Information Systems, Decision Support Systems, and Information and Management.