Avoid data warehouse project failures with these tips

We explore six key reasons for the failure of a data warehouse project and suggest ways to avoid them.

A data warehouse project is an expensive affair. Enough stress cannot be laid on the role of a good data warehouse in forecasting and decision-making processes across the enterprise. Here are ways to ensure success of the data warehouse project for your company.

1) Top management sponsorship: Even though a data warehouse project is more technology oriented, the aim is to primarily help the senior management to seek insights into the information they already possess. Thus, sponsorship from the top management, ideally the chief executive (CEO), is very important. If you look at the information hierarchy model, the top level of the pyramid is usually the CEO who is looking at the information in a dashboard format. The person at the bottom of the pyramid is a transaction guy. Hence, collecting information and trying to scope it should be done from top to bottom of the pyramid.

2) Avoid big-bang projects: A data warehousing project that aims at building the entire data warehouse of the company is bound to fail. Business intelligence (BI) projects are ever evolving. An enterprise resource planning (or any transaction system) has a start and an end. However, a data warehouse project continues till the information needs are reached and the maturity levels are acquired.

While implementing a data warehouse project, minor projects can be stepping stones. Once a small project has been accomplished, other departments can be added to a large data warehouse. At the same time, while architecting for the solution, the scope has to be open to collate (connectivity and interchangeability) all the data marts if they are interlinked.

3) Defining roles: The data warehouse project should be divided between business and technology. A third party undertakes business analysis and bridges the gap between the two. Ideally, a technical person should not do a business job and vice-versa. Thus, a third party is needed for this.

4) Concerted effort: The data warehousing project requires absolute focus. Lack of business commitment would result in unclear goals. The data warehouse project’s roadmap should be made by the senior management, with the CEO and chief information officer (CIO) working closely. The CEO may play the role of the project manager himself. The CIO plays an important role as the CEO is not concerned with the kind of data goes into the warehouse. He cares about how the data is going to come out and be presented in front of him.

5) Cleaning data: No data is clean today; data quality issues will always exist. Hence, a data quality solution should be implemented at the beginning, even before starting the data warehouse project. Another way is to clean the data as the project goes along. Ideally though, the data quality issues should be tracked at the application end. Also, the data already in the system could be cleaned using data quality tools and by stopping additional data from coming into the system.

6) Insufficient vendor support: Many implementation vendors are just body shoppers. They do not have the skill sets to implement or design a solution clearly. Thus, it becomes crucial to have a data warehouse project partner who understands the technology well and knows how to architect a solution.

Most of the vendors for data warehouse products do the same thing. Support is crucial from the vendor point of view. The vendor should also take part-ownership of the delivery.

About the author: Imran Quraishi is the vice president for BI and CEO at iDecisions Consulting FZE.

(As told to Sharon D’Souza.)

Read more on Data warehousing