This article originally appeared on the BeyeNETWORK.
One of the problems that often plagues companies is deciding who should manage the Corporate Information Factory. The business intelligence tools that are available today, combined with the computer literacy of business analysts, have evolved to the point that the business units often want to be in charge. On the other hand, tradition, corporate accountability for computing resources and economies of scale, favor the Corporate Information Technology Department having that role. To address this often politically charged issue, requires an understanding of the major components of the Corporate Information Factory.
Within the Corporate Information Factory, there are several significant data stores that are created – the “data warehouse” and the “data marts”. (For purposes of this article, the term “data warehouse” encompasses both the traditional data warehouse used for strategic analysis and the operational data store used for tactical support.)
- The data warehouse is created to provide an enterprise view of the information of interest. Therefore, development of the data access capabilities to populate the data warehouse requires integration of data from multiple sources into a flexible structure that can supply data for each unit’s information retrieval needs.
- The data marts are created to provide individualized or departmental views based on specific user group or functional requirements. Hence, development of the data delivery capabilities to populate each data mart entails extracting a subset of the data warehouse content needed to satisfy only those needs.
The distinction between these two data stores is at the heart of solving the responsibility problem.
Responsibility for the data acquisition (data warehouse)
Simply stated, delivery of an enterprise view of data requires management from an enterprise perspective. The enterprise view cannot be effectively attained by managing the effort within a single business unit. History has shown us that when a single business unit manages or dictates the requirements for a major computer system, the views of that unit prevail over the views of other business units. This approach typically leads to independent applications that need to be bridged together using interfaces or integration programs.
To get an enterprise perspective, construction of the data warehouse needs to be centrally managed. A central information technology organization is best equipped to manage this type of data store. It often meets resistance from business groups who want ownership. An effective option is to manage the number within the information technology group under the auspices of a cross-functional steering committee. This approach addresses two very important attributes of data warehouse projects – skill-set and ownership. While there may be individuals in other departments with data processing skills, the information technology organization is typically the only organization in the company that is authorized to access application systems. It has the needed experience and skill-set to tackle a complex computer application development effort that can bridge multiple source systems, technologies and locations.
On the other hand, the business units need to have a sense of ownership for the data warehouse. The suggested cross-functional steering committee with business and information technology executive membership will create an environment in which the business unit executives establish the priorities and requirements and mandate the business practices for the data warehouse. The information technology member often facilitates the steering committee and provides other members with information to help them understand the technical issues. When a business unit wants to “break the rules”, its executive, sitting on the steering committee, with input from the CIO, is in the best position to enforce adherence to policies, particularly when that executive was involved in the establishment of the policies.
The steering committee is a group of business and information technology executives with the following major responsibilities :
- Establishing the mission statement and high level policies for the Corporate Information Factory. These policies include the roles and responsibilities of the information technology group as well as those of the business units. Additionally, they include restrictions, if any, on what any group may do independently.
- Enforcing compliance with the established policies and architectural directions.
- Establishing the priorities for the data warehouse efforts. These priorities dictate the sequence of the projects that add information to the data warehouse and the scope of each project. The committee may also be responsible for securing the necessary funding.
- Resolving conflicts among business units with respect to the data warehouse content and use.
- Reviewing progress of on-going data warehouse projects and sanctioning appropriate revisions.
- Establishing quality expectations.
- Actively promoting the data warehouse as the source of choice to support analyses and decision-making.
Responsibility for data delivery (data marts and data access)
Building the data marts and providing end-user capabilities requires an understanding of how the information can be used. It also requires the technical skills to design the data marts and to provide those views by using the end-user-oriented tools. While the information technology group can provide these services, often the business units already have the needed skills. When they do, responsibility for the data marts can be undertaken by the business units.
As described below enterprises that distribute the responsibility for the data marts undertake some risks, but these risks are fully manageable.
- Risk 1: The data won’t represent the enterprise view. While this is a risk, it is largely minimized by using the data warehouse, which contains the enterprise view, as the source for all data marts. Sometimes functional areas need only a subset of the data (e.g., sales for a particular group of consumers), and the associated meta data can explain the deviations from the enterprise view.
- Risk 2: Redundant work will be performed. Without awareness of what each group is developing, multiple groups may independently develop similar capabilities. The result is a productivity loss. This risk can be mitigated by providing conformed dimensions to be used by all business groups and by providing forums for publicizing data warehouse applications.
- Risk 3: A capability developed by one group could provide benefits to other groups. Without awareness of what each group is developing, innovative applications by one group are not known, and therefore are not applied by other groups with similar needs. The result here is an opportunity loss. The mitigating action for this risk is similar to that for the second risk. By publicizing new capabilities, business groups can become aware of opportunities to leverage the work of others.
In a mature environment, the data delivery area of the Corporate Information Factory is more volatile than the data acquisition area. When the business units are responsible for data delivery, they have a feeling of being in control. The central information technology group can play an active coordination role. It can be responsible for facilitating interactions among the business units through user forums, user newsletters, etc. Making a data mart developed by one group widely accessible requires work (e.g., documentation / meta data, additional testing, etc.), and the original creator may not have an incentive to make the necessary investment. The centralized group can help, by either facilitating making the capability more publicly available, or assuming responsibility for making the capability “production grade”. It may even assume responsibility for executing the data delivery programs.
Business units and the information technology group each have a legitimate role in the development and maintenance of the Corporate Information Factory. Through careful planning, compromises can be reached to ensure that the data warehouse and data marts can each meet the stated objectives, while being sensitive to the corporate culture and business practices.
 When a data stewardship function exists, the data stewards may assume some of these responsibilities.