Watch out for BI mess

A plethora of new reporting tools flooding the market, coupled with consolidation within the business intelligence (BI) industry, will leave many enterprises facing the challenge of managing multiple ERP and BI systems, according to a panel of BI vendors.

Representatives from BI vendors gathered last week to discuss how enterprises can avoid making the same mistakes made by those that pushed data warehousing in the 1990s when rolling out their BI solution.

"I haven't seen a large company with a data warehouse that has worked," said Peter O'Donnell, lecturer, Decision Support Systems Laboratory, Monash University. "Mainly because of a mismatch between cultures that don't want to share information. So you have these large companies that are all decentralised. There's no centralisation."

The panel agreed that a boom in analytical applications on offer will present more challenges for the CIO that does not have a clear strategy in place.

"Large-scale, enterprise-wide data warehousing is very risky. BI efforts and data warehousing efforts need to fit into the organisation. But in many companies, people don't want to share information," he said. "This is not something IT can fix. Unless you've got a culture to share information, than it won't work."

Patrick Elliott, managing director, Hyperion Solutions, Australia and New Zealand added, "IT departments that just go ahead and do it account for most of the failures we've seen, because the data warehouse generally does not conform to the organisation."

A lot of the clients Hyperion deals with avoid the term data warehousing and opt for terms such as 'operational data stores'.

"The business has to drive the requirements, not IT. The CIO needs to ensure that toolsets and solutions are chosen based on their individual merit/benefit and also the role they will play in the overall business intelligence strategy," Elliott told Computerworld.

Although enterprises that struggled with data warehousing now face similar challenges rolling out a BI solution, BI vendors predict there will be a proliferation of BI tools on the desktops, which will create further complexities for enterprises.

"The CIO will not want to have to support 10 or more BI tools but would probably find themselves supporting two to four different front-end tools along with an OLAP component - for summarised and analytics - and a relational component - for transactional data," Elliott said.

O'Donnell said: "a lot of companies have several ERP solutions from separate vendors and several BI solutions. You need a BI strategy to sort it all out."

"If ERP had got it right in the beginning, then why did we need CRM?" he said. "If ERP had been the all encompassing solution than we would not have needed other systems such as CRM and an analytical layer to access and exploit these large volumes of data across the enterprise."

Going forward the ultimate BI solution will be hybrid, according to Elliott.