Analysis: Business Intelligence
How to Increase the Reliability of Your IT Infrastructure Using Predictive Analytics REGISTER >
On a cold December morning last year, Tom Lesica, the COO and CTO of NewRoads Inc., sat down in his office in a stone and glass building tucked away in the leafy New York City suburb of Greenwich, Conn., took a sip of coffee and booted his laptop. Up came his "digital dashboard," a graphical and numerical summary of the state of his business. At first glance all was well, but then he noticed a button glowing bright red. Something was wrong in distribution.
Most executives worry about their customers. Lesica worries about 200 sets of customers. His firm provides outsourced customer service and order fulfillment for direct-to-consumer activities at such companies as The Procter & Gamble Co., Avon Products Inc. and Godiva Chocolatier Inc. Given the condition of the economy, attention to detail during the holidays is critical. "We're the one throat to choke," Lesica says of his clients' expectations.
Lesica clicked through the red indicator, digging deeper into the data for its cause. There it was: Returns the day before were higher than forecasted. More clicks, and he isolated it to his distribution center in Martinsville, Va., and then to one client, a clothing retailer. Now he's on the phone with the operations manager in Martinsville and with the client, who are looking at the same numbers. In minutes, they see the problem: A yellow dress featured on the client's Web site turned out, in reality, to be another shade entirely. Changes are made on the site. Lesica's coffee is still warm. "Try to imagine yourself doing this in a spreadsheet world," he says.
What's going on here and in similar offices across the country looks like the future of business: information flowing in real time up and down the enterprise, across organizational boundaries, and out to partners, suppliers and customers. From the huge masses of data emerges both a snapshot of current operations and a basis for objectively refining the business in the future. "Any company that wants to maintain an innovative position has to have this type of interactive software, upstream and downstream," says James Brian Quinn, the William and Josephine Buchanan Professor of Management Emeritus at Dartmouth's Tuck School of Business, and the author of Innovation Explosion.
Business intelligence (BI) software or "analytics," grew up in the 1990s as companies began wondering what to do with the huge quantities of data they were accumulating from their ERP systems, call centers and the Internet. Lesica is sitting on more than a terabyte of sales transaction data, while certain financial services firms store dozens of terabytes. At some firms, Gartner Inc. reports, the data is growing by as much as 100 percent a year. Surely, the thinking went, lurking in that data is intelligence that could be usedif someone could just make sense of it. "We process 70 million to 80 million transactions a year," Lesica says, "and we can learn a heck of a lot about products or consumers or operating effectiveness from the transaction data. Let's say you place an order with one of our clients. That transaction kicks off many events: We check that it's in stock, we ship it, and then we e-mail you to let you know it's on the way. Meantime, the system is adjusting inventory, looking for back-order situations and providing reporting on business performance. One transaction can drive intelligence up and down the supply chain, from the consumer back to the manufacturers and suppliers of the product."
Lesica, who installed similar systems while serving as CIO of PepsiCo Inc. and then at J. Crew Inc., is in the early stages of rolling out the digital dashboard technology to all of NewRoads' 20 locations and 200 clients. As this Web-based technology replaces an already sophisticated reporting system using EDI, file transfer and other means, he anticipates shaving weeks off decision making and moving to a real-time, "events-based" operationallowing a client, for instance, to suspend an e-mail campaign the moment inventories disappear. He's been planning it for a year. Interfacing with 200 different client systems is one issue, but equally critical is getting everyone to agree on standard definitions and metrics. Does a phone call, for example, start the second a customer connects to the call center or when he actually begins talking to an operator?
True analytics isn't simply a matter of trusting your company's future to a screen full of clever charts and flashing lights. Can analytics ever go wrong, sending you marching confidently forwardand over a cliff? "Can computers make mistakes? Absolutely," says Mark Hurd, president of NCR Corp. and COO of its Teradata unit, an analytics database and software vendor. "Computers can make mistakes only in the context of the logic that's applied to them. I don't think you ever want to replace a human being at the other end of this who makes judgments. You have to get the humans schooled, and they need to take advantage of all the learning."