MMH Data Warehousing, Corporate Portal & e-business Applications.

Data Warehousing, Corporate Portal & e-Business Intelligence Applications

  

Mistakes to Avoid in Building Data Warehouses

Table of Contents:

Why Do Data Warehouses Fail?

Numerous surveys indicate that a high percentage of data warehouses fail to meet business objectives or are outright failures. What causes these failures? What lessons have we learned about how to avoid these failures? What can you do to ensure that your data warehousing initiative will succeed?

Some of the common sources of failure of data warehousing efforts include:

  • Failure to define the business drivers for the data warehouse
  • Use of the wrong architecture. Data warehouses are often built several times before the correct architecture is utilized
  • Population of data warehouses with "dirty" source data containing missing, inconsistent, and erroneous data values
  • Development of "stovepipe" data marts that satisfy the needs of individual business units, but do not support corporate data warehousing objectives
  • Implementation of an enterprise data warehouse as a single, large, top-down development effort
  • Failure to anticipate and overcome scalability and performance issues

These mistakes are common and difficult to avoid. A primary objective of this paper is to help you build a successful data warehouse correctly on your first try. Each of the potential causes of failure of your data warehouse is discussed below.

Failure to Define Business Drivers for the Data Warehouse

Business managers or IT managers often decide to build a data warehouse without clearly defining the business purpose of the application. Managers may be motivated to build a data warehouse due to the successful implementation of data warehouses by competitors in their industry. They may not know the specific business problems that will be addressed by the data warehouse, but they recognize that data warehouses can be used as an important tool to maintain a competitive advantage. IT managers may initiate the development of a data warehouse, in advance of a business requirement, in the hope that business analysts will recognize its usefulness to solve business problems.

However, the data warehouse is likely to fail if managers do not build the data warehouse to solve a well-defined business problem. The business problem must be specific and painful. Examples of painful business problems that can be addressed effectively by data warehouses include the following:

  • Organization is losing its competitive advantage
  • Customers are moving to competitors
  • Management has little insight and control over costs
  • Promotions are failing for unknown reasons
  • Low turnover of goods and high cost of inventory
  • Inadequate understanding of customer needs

Data warehousing is a powerful technology that can be used to solve business problems and meet strategic objectives. These objectives include improving services for customers and end users, reducing costs of business processing, increasing profitability, meeting increased global competition, responding faster to competitive challenges, providing transparent access to data, and supporting a faster, more informed decision-making process. Data warehousing can provide a competitive advantage for organizations by increasing market share through analysis of customer profiles, including scores based on spending patterns, product usage, and demographics. Additional applications include analysis of customer value and profitability across products, customer retention, behavior modification, cross selling through an integrated view of the customer, managing inventory better across geographical locations, forecasting demand patterns, and discovery of pattern and trends in the data. Data warehousing supports a single, consistent source and definition of data, with improved data accuracy and quality.

However, it is essential for business managers to identify the specific business problems that the data warehouse will address, and define a set of measurable business objectives for the data warehouse. Examples of measurable business objectives include the following:

  • Increase response to promotions from 2% to 5% through a better understanding of customer demographics
  • Save $20M by cutting attrition of top 1/3 of customers by 2%
  • Cut the expense of finding new customers from 65 - 75 cents per lead to 4.5 - 6 cents per lead (large telco)
  • Reduce the cost of fraud in payments to medical service providers by 10% in 12 months
  • Reduce the cost of service calls by 10% and reduce the cost of repair parts by 5% in 18 months
  • Reduce reject rate in manufacturing from 3.5% to 2%

An important objective of the development effort for the data warehouse is to minimize development risk and provide a rapid Return On Investment. The specific goal, addressed in this paper, is to provide a tangible Return on Investment in 90 to 120 days after initiation of the development effort. The development approach recommended in the paper is to first design the overall enterprise data warehouse architecture on paper, and then implement the data warehouse bottom-up, one business area at a time.

[ TOP ]

 
 
 

©2001 Mimno, Myers & Holum