I’m here on the business intelligence track at the Burton Group’s Catalyst Conference, trying to sort out the old BI from the new. As you might expect, there is a lot of talk about predictive analytics and complex event processing. The data warehousing of the past is done! Accessing “data on the fly” so the business can nimbly navigate the new normal is in. Yesterday’s theme was what IT needs to do to deliver business insight, not just business intelligence.
At the first session I learned that the old BI — BI at the dawn of computers — was there to help companies automate and become more efficient by taking the human factor out. The side benefit was that it reduced the tasks that humans had to think about (presumably so they could think about even harder questions). The approach was highly successful, but it did not anticipate the massive amount of data that businesses accumulate. The automation paradigm has run its course. The focus of the new BI should not be on removing the human to gain efficiency — those efficiencies have been realized — but getting the human back in the game. And not by handing the business another static (yawn) report that itemizes or narrowly analyzes data. The business doesn’t want to wait for answers from IT. The new BI is not about delivering answers at all, but about building architectures and tools that allow individuals to discover the salient pieces of the data. IT should focus on finding more powerful ways to assemble data to help discover why something happened, not just what happened. That was one track I heard.
By the next session, I was hearing that BI needs to automate more, by using complex event processing (CEP) tools to correlate tons of information that will allow businesses to take real-time automated action. Instead of getting out of the way of the business, IT needs “to lead the way” on complex event processing, according to the analyst. Some industries are already deep in CEP. Casinos do it well. The airlines are getting better at it. Of course, the financial services industry nearly brought the world economy to an end, partly by doing this. But I didn’t hear much about risk on the BI track. Or about privacy issues related to the stores of personal data required to turn complex event processing into something that helps a business improve customer service.
When I asked about the danger of taking automated action based on potentially bad data — on the kind of scale, mind you, that we saw in the financial services industry — I heard about “feedback” loops that adjust actions according to mistakes. Consider, I heard, how complex event processing can reduce risk by correlating data to instantly alert a theater of a fire and activate safety mechanisms, thus minimizing the loss of life. Unless the data is wrong, I was thinking, and the automated response causes a needless stampede to the exits.
One aspect of BI that just about everybody seemed to agree on: Data is precious. Or, as I heard at today’s session about fungible and nonfungible data, “Data is not the sawdust of processes.” More tomorrow, about the difference between fungible and nonfungible data. (P.S.: It’s not as clear-cut as you might think.)
Let us know what you think about this post; email Linda Tucci, Senior News Writer.