While true empowerment includes disseminating business intelligence to as wide an audience as possible within the enterprise, companies should avoid dumbing down analytics, or going for a "black box" business intelligence approach.
So says Bill Hoggarth, managing director of SAS Institute in SA, who points out that enabling a far greater percentage of employees to leverage business intelligence analytics does not mean sacrificing standards. "It can be achieved not by adjusting the technology, but by adding delivery mechanisms on top of it."
Thus, companies still need their 'rocket scientists' to create the initial statistical models, but many more employees - without PhDs - are able to use business intelligence.
"Analytics must be disseminated to everyone within the company that can use them," says Hoggarth. "In the past, all business intelligence vendors were focused on the start of the information supply chain - that is, getting access to data. Then they all focused on storing the data, and being able to analyse it, before focusing finally on having tools to analyse the data differently."
The big focus today for SAS, leader in business intelligence, is how to get this breadth of information to the wider audience.
"We believe that 80% of an organisation's enterprise needs are business intelligence," says Hoggarth. "There is a worldwide shortage of business analysts, so we have had to empower domain experts and information consumers to do some analysis themselves."
The latest version of the world's leading business intelligence software, SAS Version 9.1, delivers a breadth of interfaces that empowers the vast majority of an enterprise to use business intelligence, instead of a very small percentage.
"We have not dumbed down the analytics to achieve this. We have, however, ensured that when they get used, people who know how to use them are creating a safe environment for the rest of the knowledge base.
"Dumbing down and black-boxing analytics are very dangerous, because models have to be retrained to notice the subtleties in the data," he warns.
"If you build a statistical model, it is simply a bunch of nodes hooked together that do not particularly know anything - yet. You need to run data through them, which could creates a decision tree, for example, one type of data mining model.
"The model must be trained, and then retrained from time to time because the information in the data changes. Adjustments must be made to the model, based on this new information, to keep it fresh or even make it better at times."
Share