Strategy needed for business information

Management of many organisations are calling for strategic business information from their ICT systems. But they have no real business information strategy, nor a dedicated team to create one, says consultant Roy Elgar.

Currently with Teradata, Elgar has consulted on data warehousing and strategic business information with companies worldwide that have some of the largest and fastest-moving collections of data.

Bharti India's largest cellphone network company, for example loads 1.2 billion call records a day, he says.

Improved business information has turned up repeatedly for the past three years in industry analysis of the top priorities signalled by the business to CIOs in the US and Europe.

On a smaller scale, New Zealand companies are faced with the same need, he told an audience of Computer Society members earlier this month

Yet, there is little in the way of a truly strategic approach; what often emerges is a set of departmental point solutions.

Setting up a centre of expertise - even if only one person - simply to manage BI is critical, he says. A specialist staff member or team can take the larger view.

A company-wide data warehousing exercise analysing the needs of all departments in concert will have important synergies, Elgar says.

Most items of data will be used to answer the business information questions of more than one department, yet the departments often collect and maintain those items separately in their own datamarts, with massive duplication and a risk of inconsistency.

One of the most prevalent errors in setting up a BI system, Elgar says, is to confuse technological elements with pure logical data questions.

There are five separate layers to the exercise: the business goals; applications that provide the information; a logical data model (with no reference to technology) on which the applications work, a set of "views" which offer users a role-dependent "view" of the overall logical model and a physical data model that implements the logical model efficiently.