Ever since “business intelligence” entered the lexicon a couple of decades ago, the technology has been mainly confined to use by research professionals–usually statistics zealots–who crunched, sliced and diced data in their offices to spot trends in markets and operations. They provided decision-makers with good insights into business trends–last quarter’s business trends, that is.
Unisphere Research, the research arm of DBTA, has conducted surveys that confirm the existence this BI “elitism.” For example, research conducted in cooperation with the Oracle Applications Users Group (OAUG) found that at the vast majority of companies, 90 percent, only about 10 percent of employees actually have access to or use their installed BI solutions. (Not counting spreadsheets, mind you.) Most of these companies want to extend BI to more employees, but it would be probably counter-productive to have front-line employees and managers doing their own ad hoc drilling, slicing, and dicing of data.
What is needed instead is a variation of BI that exists more behind the scenes, operating on consistent, pre-established rules across the enterprise, serving up insights almost automatically when and where needed. BI should be behind the everyday applications that employees use to do their jobs, both in a tactical as well as strategic sense. Some call it “pervasive BI,” others call it “operational BI,” or “embedded BI.” Whatever it is, it all leads to one thing–BI functions best when built into day-to-day transactional applications.
Pervasive BI deals with up-to-the-minute transactions and interactions and provides actionable insights on what decisions need to be made at that moment. Should a call center representative grant the customer on the line a lower interest rate on a credit card to keep her business? How should an airline representative at the gate reroute that valuable customer who just missed his connection? What is the medical history of this patient who was just brought in? Which truck should be unloaded first at the congested loading dock? A pervasive BI system can deliver insights based on pre-programmed rules and input from other processes.
Vendors are stepping up to the challenge. I recently had the opportunity to sit down with Paul Grabscheid, vice president, strategic planning for InterSystems, at the company’s Cambridge, Mass., headquarters, for a glimpse into this new direction for BI.
Pervasive BI provides just-in-time insights when and where they are needed–and may be based on a blend of historical data coming from the warehouse and real-time data coming right in from a production system. Grabscheid said his company, like most others, up until now have been “focused in the kind of run-the-business transaction processing part of what organizations do. If you look at where our systems end up, or technology ends up, what people build with them, its those core, repeated, do it over and over again, its got to work all the time kinds of systems.”
Many enterprises have hit the limits as far as leveraging BI capabilities, and are looking for tools to help managing realities on the ground. “What began to change our view of the world a couple of years ago,” Grabscheid explained, “was more and more customers and application partners were saying to us that ‘the data warehouse is all good, but we’ve got some other analytic needs that are more operational in nature, that really belong inside our transaction processing run the business applications, that really need access to up the minute information, not stuff we extracted last week.’”
There have been notable advances in making BI data timelier–from months and weeks old to days and hours old. In pervasive BI, data isn’t necessarily from three seconds ago–rather, analytical historical data is blended with real-time transactional data for automatic analysis and delivery of recommended actions to front-line employees and managers.
Other data management vendors are focusing on pervasive BI as well. I’ve been following the discussions that James Markarian, CTO of Informatica, and Stephen Brobst, CTO of Teradata, have been having about pervasive BI in recent Webcasts and white papers. Brobst observed in a Webcast that current traditional BI tools are not made for pervasive BI. He noted that pervasive BI is “not going to happen by deploying traditional BI tools to the people on the front line.” Rather, pervasive BI will gradually become part of operations delivered through “decisioning services” made available to the entire enterprise, both inside and outside the firewall.
In addition, these decisioning services can be brought online in an incremental fashion – versus some type of big migration project, Markarian added. “Pervasive BI doesn’t mean an immediate wholesale change to the way that we do everything. It’s an architecture and framework.”
However, the challenge will be working through many of today’s enterprise data management solutions and platforms, which are configured to operate within silos serving specific processes or functions. “In pervasive business intelligence, you’re looking for something a lot more holistic, where you’re combining information with different latencies,” Markarian said. “This is the real-time index with the pervasive view. We need to create a single architecture that can handle all workloads.” The way to do accomplish this, Markarian said, is through building out an integration or data services layer that will support and rationalize data coming in from varied sources with various degrees of latency.