Is it possible business intelligence solutions are too sexy for their own good? Yes, according to one Monash University lecturer.
Peter O’Donnell has been working as a business intelligence applications lecturer at the university since 1990. In that time he has focused on testing things people take for granted in the BI space – including sex appeal.
"As a collective BI industry, everyone has the rhetoric that it is about visualisation and letting people see their data; it's a grand dream that we put this slice and dice interface in front of people," he said. "That is the way vendors do their demos and somehow or other people will drop, click and drag to get to the relevant number and their day is better and the problem solved."
But one of O'Donnell's concerns is all the main vendors' products are similar with the same style of interface. And it's an interface that O'Donnell contends has contributed to low utilisation rates – often as low as single digits he claimed – despite strong sales.
According to analyst firm Gartner, the Australian BI market hit $US211 million in 2008 for growth of 11 per cent. BI applications were also ranked the top technology priority in 2009 for the fourth year in a row in Gartner’s annual survey of CIOs.
Globally the appetite for BI platforms, analytic applications and performance management software in 2008 increased 21.7 per cent on the previous year, from $US7.2 billion to $US8.8 billion.
Yet, according to O'Donnell's research, many employees don't get the full benefit out of BI applications. People who acquire knowledge and learn visually or kinestheticly (by doing) are relatively comfortable with fancy graphical representations of data in BI solutions, but those who learn aurally or by reading aren't.
"A lot of the tools are very sexy. You can do 3D pie charts and donut charts; in theoretical terms it is known as the data to ink ratio. If you have embellishments and 3D effects you are using a lot of ink for only a small amount of data," he said.
"In a sample of 2 groups of 20 people, we gave the same data to each group. To one of them we gave them plain, boring looking charts that adhere to the guidelines of keeping the data to ink ratio close to one. The other one we gave them sexy stuff. One group would get a bar chart the other a pie chart. One would get a bar char the next would get a bar chart with 3D effects. We'd ask them to analyse very simple questions – i.e. compare this data, which one is bigger, which one is smaller, etc. The results were generally awful. There was something like a 62 per cent probability of getting the questions right if you got the plain charts. If you got the sexy ones it was like 47 per cent."
While O'Donnell freely acknowledges it is hard to generalise from such a small sample he concludes there are significant problems with BI solutions using sexy graphics that are trying to support decision making processes.
"When all the vendors have tools that look the same, we may be supporting one group of people with one particular cognitive style and missing some others," he said.
One of the reasons for the disconnect between strong sales and poor utilisation rates is there hasn't been enough focus on employing developers that have advanced interface skills. O'Donnell's research team has been observing the Australian job advertisements for BI positions. The findings appear to back up his position.
"Everyone says the front end is important but all the job ads are for data modelling skills or data transformation skills. I have seen in the last six months, and there are 500 or 600 jobs in Australia at any one time, one job mention anything about usability. They generally have got something on, must have business objects or must have used COGNOS or that kind of stuff – clearly front end technologies. But nobody is going beyond the tool set and saying they must know about visualisation or usability."
In short, BI vendors may be capable of creating sexy hot graphics to present data, but they have failed to engage users. For O'Donnell providing greater choice, and in particular plain Jane options for graphics, would help boost the utilisation and success rates.
"I don't think people are using the slice of dices to get the wrong data," he said. "I think they just give up."