A clean approach to BI

Companies that view business intelligence (BI) as a back office function are trailing behind their competitors that make this function a part of their everyday business practice.

Tan Bee Kee, practice principal, information management, HP services, consulting and integration, HP Asia-Pacific and Japan, said unlike the BI scene some years ago, business analytics is now seen by many companies as a critical element of business strategy.

The trend toward "operational BI", which Tan defined as the integration of BI into business operations, ties business decisions tightly to the interpretation of analytical data.

"BI has gone mainstream. It sits at the core of mission-critical applications, and it is not just for analysts or backend personnel, but in the forefront with business decision-makers," she said, speaking at the SAS Forum 2008 event held in Singapore Tuesday.

Companies are integrating their BI set ups to major backend systems like their ERP (enterprise resource planning), CRM (customer relationship management) and SCM (supply chain management) systems too, added Tan.

As a result, downtime of the BI systems "is not tolerated" by businesses, placing even greater emphasis on the running of the systems, she said.

Data quality is the business' responsibility
Besides upkeeping the systems, a greater challenge lies in ensuring data quality.

"Data quality has become part of the overall business agenda...it can't just be a one-department responsibility" she said, adding that the duty of ensuring data cleanliness has moved from the CIO's shoulders because of the breadth and pervasiveness of mined data that goes into a BI system for processing.

This data includes unstructured data, especially during acquisitions, that must be transformed into accessible, usable information, she said.

Mark Perez, vice president and head of branch banking support center at the Metropolitan Bank and Trust Company, said in his speech at the same event, that companies need to find a way to balance "real" data against that in their databases.

Explaining "real" data, Perez said it refers to information that comes from sales people in the field gathering customer feedback. This data needs to be taken into account when running business analytics, but it sometimes causes problems for BI staff when it contradicts data that is already in the database, which comes from other sources such as tracking customer behavior.

Nonetheless, the trick is to balance both sets of data within the BI system to find "which road is safe" to take, and for the marketing division to continually refine the questions they ask in data gathering, noted Perez.

Tan said another common source of unstructured data comes from Excel spreadsheets. The propagation of such files that are passed around a company creates "shadow databases" which are not synced to the master database and often result in incongruent information.

"But can you ban Excel? Of course not. Almost every worker needs it as the common user interface," said Tan.

As such, companies need to draft policies supporting Excel use, so that workers and IT can prevent the mismatching of data that might occur otherwise. "You can use Excel as a display interface, but workers must always go back to the main source of data to update," she said.