Identify BI users' needs

By Vicky Burger, ITWeb portals content / relationship manager

Johannesburg, 06 Nov 2007

Business users instinctively understand data organised in a dimensional format, says business intelligence (BI) guru Dr Ralph Kimball.

In a career spanning 35 years, Kimball, founder of the Kimball Group, says: "Fortunately, this dimensional format is easy for database systems to process at very high speed."

Kimball will discuss this dimensional approach to BI on 13 November when he addresses delegates at the one-day Straight talk on Practical BI Techniques for Proven Results seminar, to be hosted by Sybase SA's BI Practice, in association with ITWeb.

"The data warehouse is the platform for business intelligence," says Kimball. It has the responsibility to extract, clean, and prepare the data to be in the simplest and most accessible format possible.

If this is done, he adds, all BI tools will be able to present the data to the end-user quickly through simple user interfaces.

Pitfalls

Kimball states that by far the biggest mistake companies make in building an effective data warehouse is to inadequately identify the needs of the business users.

These needs are the key performance indicators (KPIs) that are used by the business users to make decisions, he explains. "We should remember the original definition of the data warehouse is a decision support system."

In addition, not adequately profiling the data is a mistake often made by companies, reveals Kimball. "Before the data warehouse project begins, the potential data sources need to be carefully analysed to determine if the data is of a sufficiently high quality."

It must be determined if the data can support the intended mission; a late revelation that the data cannot support the mission can be costly, he adds.

Another mistake is letting content, rather than time, drive the schedule. "The data warehouse team should not sign up to a specific set of reports without realising these reports require many separate data feeds from original sources," he says.

The delays that result from building the data extract pipeline from the original sources can be fatal to the project, he notes.

Procedures to implement

Most important, says Kimball, is to have a true and accurate understanding of the business requirements that describe how the organisation makes decisions.

Executive management support for "an analytic culture that esteems decision-making with help from the numbers" is important, he states. In addition, the executive management should also provide useful guidance to the data warehouse team for implementing security, privacy and compliance.

A cross-departmental effort to establish common descriptors for the major entities in the organisation, such as customer, product, location and time should exist, he says. There needs to be assistance from the finance department to set guidelines for allocating costs to the components of revenue so profitability can be calculated in the data warehouse.

Both business and IT needs to support and partner with each other, he says. The business needs to have a full time sponsor at a senior level, to make sure it is responsive to the needs of the IT designers and implementers.

"The IT team needs to live in the end-user environment permanently."