First up against the wall

Old ways of thinking about BI, and old ways of doing it, will be first up against the wall when the revolution comes, if some of the newer players in the market have any say in the matter.

“Traditional BI is broken - something new is desperately needed,” says Qlikview head Davide Hanan. “According to research firm IDC, implementations succeed only 31% of the time, at best. Yet in today's economy, the need for reliable, accessible information has never been greater.”

Riaaz Jeena, GijimaAst BI business unit manager, agrees: “When the CXO or management go on a BI initiative, it always takes a long time to deliver. We find many clients give 18 months to three years to roll out. By the time it is delivered, the questions asked upfront are irrelevant. The market has changed, the company has changed.”

He recommends taking a bite-size approach to BI. “It's critical to get something out there as fast as you can and try to remove the 'takes too long, costs too much element'...”

Not every organisation is ready for radical change, though.

Says Martin Rennhackkamp, COO of PBT: “There's no drive from the top. Executives still get monthly report-backs and that's how they manage the business, not by looking at dials and switches online as things happen.

“Even if you have sponsorship, it's no use without the insight and understanding to see it's not delivering to the extent it can and isn't as agile as it should be.

“Another misconception of the BI fraternity is that if the information isn't accurate, it's a failure,” says Mervyn Mooi, director of KID. “Half of our clients reported things they didn't expect. It opened a can of worms. A BI system or data warehouse is supposed to show the bad in an operation, but often the executives want to hide that.

“We had one situation where we had a debate [about data], so we checked and retested, only to find out that for five years, managers were reporting incorrect information. This, after we were told we were failures and not getting the requirements right,” he comments.

As technology has evolved, BI has moved beyond what even some vendors think it is, in some quarters anyhow.

Says Hanan: “The need for quality information is helping to accelerate the move towards newer BI technology that offers quick time to value - typically the first results are seen within weeks - and much lower total cost of ownership. Old BI systems take anything from five to 18 months to implement, cost millions of rands to maintain and only work the way they are supposed to about a third of the time. New BI takes around 10 days to implement, costs a lot less and is successful about 99% of the time.

“This is possible,” he says, “because the underlying technology has improved beyond recognition, with newer software able to take advantage of massive increases in processing power and available memory. In-memory processing allows almost infinitely flexible navigation of the data, with no need to save resources by creating pre-structured data cubes. Gartner estimates that by 2010, 70% of Global 1 000 organisations will load detailed data into memory as their primary method to optimise BI application performance.”

Says Knowledge Factory CTO Peter Oeschger: “We did an RCI project recently. Before, its clients were subjected to a barrage of communications from the company. It was sending out 200 000 SMSes. This was BI for them. They had a team of highly paid BI consultants - seven figures in some cases - churning marketing lists out of an SAS back-end and getting figures in terms of how they were doing in terms of global budget.

“We gave them an SQL geospatial database. Geo-coding let them target the client a lot better in terms of 'the client lives here and holidays there'. They got better hit rates.

“The next problem was that once they had the customer on the line, wanting to book timeshare, it would take about four minutes to walk through the menu system. And if the resort didn't have what they wanted, they had to go back and do it over.

“Spatialising actual resorts and membership and making stock available visually took booking times down from five minutes to about a minute, 23 seconds.”

Aside from better customer service, this meant RSI needed fewer agents too.

“Visually exposing the data took two days, and the system went live in two weeks. RSI had 500 people and is now down to 200 by working smart. The customer gets better service and the company has reduced its salary bill by seven bar.

“It's a bitter pill to swallow,” he adds, “but it's part of the decisions BI needs be driving these days.”