By using website you agree to our use of cookies as described in our cookie policy. Learn More


LESSON - Successful Business Intelligence Implementations Start with Good Data Model Design

By Donna Burbank, Director of Product Marketing, CA ERwin Modeling

In today’s information-driven economy, organizations rely on business intelligence (BI) applications to make strategic business decisions. BI applications allow business users to create their own reports through an intuitive user interface, removing reliance on IT and putting the power of “self-service” into the hands of business users. The self-service approach is a driver in the increased popularity of BI applications, and a key justification for the return on investment (ROI) achieved.

A BI report is only as good as the database on which it is built; however, many business users are not involved in the design or population of the databases that drive their BI applications. Instead, database design and implementation are often performed by data professionals with specialized knowledge. This specialization is understandable, given the highly technical nature of database systems, but databases must be built to implement the needs of the business.

Start with a Good Design and Involve the Business

A common communication medium between IT and the business is a data model, which is typically built at various levels to suit the intended audience and purpose. For example, a logical or conceptual data model generally defines core business terms and rules, and a physical data model is used to create the database structures that are optimized to store and retrieve information. Usually, information about the business is gathered into the logical model by data modelers and architects through a series of interviews with business people.

This business-level analysis should not be overlooked—though the physical database design is important, the business rules and definitions are critical to the success of the business intelligence implementation. If a business user creates a report of product sales, the definition of “product” must be clear—does it mean finished products, or does it include raw materials? These core business terms need to be defined in the business-level data model.

Create an Inventory of Information Assets

In addition to understanding business requirements, the physical analysis of the database infrastructure is equally critical. With the massive volume of data in most organizations today, it is a challenge to understand what information exists and where it is stored. Data models provide a graphical road map of information assets by reverse engineering databases to analyze their structure and interrelationships.

Product information, for example, may be stored in multiple databases, on diverse platforms, and in disparate geographical regions. Data might be stored in various formats and use different terminology. The process of harmonizing information into a single, consistent design is facilitated by the graphical nature of a data model. Like a graphical road map that helps guide you through city streets, a graphical data model provides a map of your information assets and helps enforce standards so that information can be easily consolidated.

Get the Word Out

Once an inventory is built and business definitions are created, make sure to publish the information in a format that is easily accessible to multiple roles across the organization. For example, since business users may not have access to a data modeling tool, publish the models onto the Web for this audience. Even though graphic data models are intuitive, consider publishing model definitions in other formats such as spreadsheets, wikis, or even Word documents.

Although the analysis should be done in a robust modeling tool, the publication should be flexible to reach all audiences. Remember, the success of many BI applications was in their user-friendly nature. Make the data models that define the information just as accessible and user-friendly.


The growth in BI reporting has increased the need for robust analysis on the back end to ensure reports are built on correct information. To support the front-end reports, the back-end analysis requires both a careful inventory of database assets and an interactive discussion with the business to make sure data is defined correctly. Once this analysis has been done using data models, it is important that information is shared with the end consumers through intuitive, “self-service” publications.

For a free white paper on this topic from CA ERwin Modeling, click here and choose the title “The Benefits of Data Modeling in Business Intelligence.” For more free white papers, click here.

TDWI Membership

Get immediate access to training discounts, video library, BI Teams, Skills, Budget Report, and more

Individual, Student, and Team memberships available.