A surprisingly high 87% of BI initiatives fail to live up to expectations, according to David Waddington, founder of Tyson Consulting and keynote speaker at Information Age’s Business Intelligence 08 conference.

It was a provocative statement to make in front of a packed audience of BI vendors, implementers and CIOs. But rather than scorn, it was met with a sea of nodding heads. And that wasn’t the only BI metric Waddington had to share.

“The forecast for the BI market is $7 billion,” he said, “but 27% of business leaders say the data [used in their BI systems] is inconsistent, and 47% think their staff have made poor decisions because they did not have access to the right data.

Furthermore, he noted that only “36% of CIOs believe management is using the right information to make decisions”, and “less than 5% of business leaders are satisfied with data integration and analysis”. In other words, “there’s a lot of very unhappy people out there”.

Waddington doesn’t blame BI vendors or poor software, suggesting the cause is much deeper: “We’ve all been there during a demonstration where a BI tool worked very nicely with cleansed data, and we get swept away by the design of the front end… But it’s not the tool, it’s the systems and processes and the data behind the program.”

With his experience helping companies struggling with their BI implementations, Waddington has some frightening stories of BI gone wrong. “One company in France was reporting monthly results back to headquarters in England, converting francs into pounds. The only problem: headquarters was converting them again. It took them three years to discover it was going on,” he explained.

“Another company had 13 different definitions of ‘net proceeds of sale’. They were adding pears and getting oranges.”

Such problems might be easy to solve in a small company – “you can just get new code from Fred at the other end of the corridor” – but can be expensive and confusing for larger organisations.

The modern business environment is very complex, Waddington argued: “There is a whole series of quite complex processes and pressures, including supply lines, production facilities, stock markets, accounts and globalisation. The traditional barriers of time and geography no longer exist, forcing businesses all over the world to compete, and the Internet has made the speed of change dramatically faster.”

In the midst of this, a large corporation can seem a lumbering beast, looking to BI to make sense of its environment. But if BI is failing to deliver, “how can we have the flexibility and agility to react fast to change and development?” Waddington asked.

Business, not IT

A BI tool is often an IT initiative, he said – an approach that can actively work against making the project effective: “Often, by the time IT gets to the stage of delivering a report, the world has moved on and business is no longer interested. You are constantly moving just to stand still.”

One solution is to recruit an executive sponsor, “preferably someone from C-level: the CFO or even the CEO”, and ensure business is involved in managing the project even after it is established.

“The sponsor needs to be warned though that it will be hard, and needs to stand up there and make decisions when it gets rough – and it will,” he added.

Most large businesses deal with a range of different vendors and programs, producing data in a staggering variety of formats that need to be taken into account by a BI tool for it to provide accurate information.

“Why is consistency so difficult?” Waddington asked. “We need standardisation and convergence between different products.” A common pitfall for BI vendors is asking users what they need. “Ask them what they want and they’ll say ‘everything’,” said Waddington. The result is a set of BI tools with “a mesmerising range of functions that only 10% of users are ever going to need”.

“These guys have a day job; they don’t want to learn a new tool when they are only going to use 10% of the functionality,” he said. A more effective approach, he suggested, was for vendors to work with users “and see what functions they use on a daily basis”.

For BI to succeed it must take an “information engine” approach – generating confidence in the data it produces, keeping up with change and allowing federation, Waddington said.

“In the real world of business we need a simpler and different approach.”