Is It Really Time for Real-Time?By Joshua Weinberger | Posted 2003-03-01 Email Print
Re-Thinking HR: What Every CIO Needs to Know About Tomorrow's Workforce REGISTER >
Like many phrases that capture the technology industry's attention, "real-time analytics" and "the real-time enterprise" have the ring of something desirable. But just because the words come trippingly off the tongue doesn't mean t
Begin with the obvious conundrum: How "real" is "real time" supposed to be? Any data request will return real-time results—that is, results that reflect the data available at the time the data warehouse is "hit." But is the data still real by the time the analysis is complete? Is the outcome affected if you access the data at an inopportune moment? Does the difference even matter to whatever it is you're analyzing?
The biggest hurdle may be overcoming the notion that real-time information is the right goal. As far as AMR Research Vice President John Hagerty is concerned, "being 'near-real-time' is good enough." Forrester Research Director Laurie Orlov says the standard should be "the right-time enterprise." And IDC Research Director Bob Blumstein simply says that "what passes for real time is and should be 'the time that it takes to do something without interruption.'"
They may all be right, to some degree.
Ask Kevin Smith. As the manager of customer-relationship-management strategies at the regional travel company AAA Carolinas, he'll tell you that AAA's not living the real-time life—yet.
"That's not to say that real-time analytics are not needed," he says. "For example, in the event of an ice storm, local media are in constant contact requesting the number of accidents, the number of tows, etc.
"The bottom line," Smith adds, "is cost—and not just the cost of the product," meaning the software needed to collect and deliver information instantly. Until recently, "there has been no 'out-of-the-box' [software package] that could be installed relatively easily."
That may be about to change.
Over the last 18 months, more than a dozen vendors have come to market with applications aiming to make the real-time enterprise a reality. These in- clude smaller data-warehouse firms strong in business analytics (Informatica, Sagent); private or financially struggling vendors who specialize in reporting tools (Crystal Decisions, MicroStrategy, Brio); pure-play business-intelligence companies (SAS, Business Objects, Cognos, Hyperion); enterprise resource planning (ERP) vendors trying to nudge their way into the field (Siebel, PeopleSoft, Oracle); and even a handful of customer management software companies (E.piphany, Kana) who see that niche as the real-time enterprise's most natural fit.
With all these players, consolida- tion seems inevitable—and may already have begun. Business Objects snapped up Acta last year, and Cognos spent $160 million for enterprise-performance software maker Adaytum.
As the speed of information de- livery increases, the importance of keeping clean, high-quality data intensifies. As a recent report from consultancy Aberdeen Group notes, "the faster tempo that real-time analytics offers can be a two-edged [sword]. Instead of gaining rapid insight, enterprises may be making faster mistakes—all because the data on which they are basing a decision may be faulty."
Some users think they can avoid that pitfall. For example, credit agency Fair, Isaac scores the quality of data returned in response to real-time queries for information. Insurance company Hartford Steam Boiler, on the other hand, simply avoids the rush altogether. Senior Architect Ed Gillis says real-time analytics may have "captured the imagination of a lot of trade rags and senior management" but only in the faint hope that they might finally realize "the true benefit they've been trying to get out of their ERP implementations over the last few years."
So companies that want real-time answers often think the easiest way to get them is to call on the vendor they're already using for enterprise planning. After all, as Bonnie Clinton, national manager of financial systems and strategy at Toyota Motor Sales, points out, "once you implement an ERP [system], you're pretty much married to it." But Meta Group Vice President Ruediger Spies contends that that structure may be limiting. Reports use "data from the operational systems that don't hold a lot of history." True analysis requires that history, plus input from and comparison to outside sources. As it stands, most companies are still just talking about real-time analytics. "There's a credibility gap between what companies claim to be doing, and what they're actually doing," says Forrester's Orlov.
Successfully funding and deploy- ing systems that deliver real-time information will depend, of course, on a real need. WellPoint's Moore, for example, has seen instant access lead to a "paradigm shift" in how his users interact with his firm's human-resource systems. By providing up-to-date information on company benefits, WellPoint is able to realize significant savings.
Todd Seeberger, data warehouse analyst at Capital District Physicians' Health Plan in Albany, N.Y., sees a similar need. He says, "There are certain customers that need their medical information immediately to base their decisions on."
But, AMR's Hagerty asks, "Is that analytics? Or is that a business process?" That's a question that needs to be answered, maybe not in real time—but sooner rather than later.