For Canadian paper-mill operator Abitibi Consolidated Inc., adapting to constantly fluctuating electricity costs--the result of last year's deregulation of the Ontario power industry--can mean the difference between operating its five Ontario plants at a profit or a loss.
To smooth out costs, Abitibi has implemented a real-time system that constantly monitors energy consumption at its newsprint plants and every five minutes compares that data to electricity prices posted on the Web site of the Independent Electrical Market Operator, which oversees Ontario's wholesale electricity market. "We get readings every minute about the energy consumption at the mills," says Guy Roussel, senior analyst for IT planning. If energy prices get too high, production can be cut back or shut down.
Abitibi's use of real-time business intelligence gives it an edge. But some companies, such as roofing materials manufacturer GAF Materials Corp., still aren't sold on the need for that kind of capability. GAF collects production data from its 25 plants and replicates it each night to a database at its headquarters, where it's used for cost-management planning and reporting. With the IT systems it has in place, GAF could collect production data every few minutes. But to what end? "I don't know if any business would be able to make decisions on data that comes in that fast," says Rick Stevenson, director of business-intelligence and supply-chain systems.
There has been a good deal of buzz lately--both pro and con--among business-technology managers and IT vendors about real-time business intelligence. The term "real time" is a bit of a misnomer, however, since it's been used to characterize both the updating and analysis of data every hour, every few minutes, or instantly. In some cases, real-time business intelligence taps directly into operational systems like supply-chain applications or securities-trading systems--a monumental change from updating data warehouses weekly or monthly. Any way you look at it, real-time business intelligence is about keeping pace in the accelerating world of business.
But it's not for every company and certainly not for every application. For traditional business-intelligence applications, such as analyzing sales data to detect trends or profile customers, investing IT resources to assemble a complex real-time business-intelligence system can be a waste. A business analyst trying to discern seasonal sales patterns, for example, doesn't need up-to-the-second data.
Yet as businesses change, so could views about the value of real-time intelligence. When IT research firm Gartner asked 540 companies in the United States and Europe last year how current data needs to be for analysis, just 11% said data should be updated instantaneously and continuously. But those same companies, anticipating continued acceleration, expect by 2006 that 29% of the data they use for analysis will need to be instantaneous and continuous. In contrast, the need to update data monthly for business-intelligence applications will drop from 17% to 3% during that time.
Growing demand for real-time business intelligence correlates to the shift in how that information is being used. Traditional uses have been for strategic analysis and planning carried out by top management, like sales forecasting, marketing-campaign analysis, and financial consolidation. Since these functions relate to longer-term planning and strategies, there's no compelling reason for real-time analysis.
The sweet spot for real-time business analysis is on the factory floor, at the reservation desk, or in the customer-service center, where operational managers and staff-level employees use it as a tactical tool. "The more real time you get, the more tactical you are," says Philip Russom, an analyst at Forrester Research.
That's a view shared by Teradata, a data warehouse vendor and division of NCR Corp. "Real-time business intelligence derives maximum value when you push it out to the front lines of your organization," says Teradata's chief technology officer, Stephen Brobst. While Teradata's products once epitomized the traditional batch-loaded, non-real-time data warehouse, today its system's "active data warehousing" architecture makes it easier to build data warehouses that support both strategic and tactical data analysis.
Continental Airlines Inc. uses Teradata's technology at its massive data warehouse, which stores information on all aspects of the airline's business, including passenger reservations, aircraft maintenance and parts inventories, and flight delays. Using middleware like IBM's WebSphere MQ and Informatica Corp.'s Striva, transactional data is captured and loaded into the data warehouse within seconds, says data warehousing director Alicia Acebo.
Continental has designed its data warehouse so that it can be used for both real-time alerts and long-range strategic analysis. It warns reservations clerks of any incoming flight delays, for example, and identifies passengers designated as most-valuable customers, who can then be given priority on alternative flights if it appears they'll miss their scheduled connections. Business analysts also can use Continental's data warehouse to study historical data to detect customer trends.
The most visible use of real-time business intelligence is business-activity monitoring. The monitoring combines data collection with process- and workflow-management capabilities to monitor streaming data from operational systems to detect "business events," such as production-line problems, spikes in customer complaints, and diminishing stock on a retailer's shelf. Key performance metrics and alerts are delivered through dashboardlike interfaces on managers' desktop computers or through mobile workers' handheld devices, pagers, and cell phones.
In the last year, there has been a wave of business-activity-monitoring software products from startup vendors such as Celequest, First Rain, and Iteration Software, as well as from Tibco Software, which acquired startup Praja last year. "Business-activity monitoring takes business intelligence to the next stage of evolution by making it more actionable," says Celequest Corp. CEO Diaz Nesamoney.
Silver Line Building Products Corp., a vinyl-window manufacturer, is installing Iteration's business-activity-monitoring software to notify shipping managers at the company's North Brunswick, N.J., plant if an order won't be ready for delivery. The Iteration Real-Time Reporting Suite, when up and running this month, will tie into the plant's manufacturing system and track custom-made windows as they move down the production line. For example, if the system determines two hours before shipping time that the order may be late, it sends an alert to BlackBerry devices carried by shipping managers, who can adjust truck-loading schedules. The ultimate goal is to improve customer service to Silver Line's distributors, says Dan Lyons, information systems VP.
Abitibi uses OSIsoft Inc.'s real-time performance-management software to monitor the cost of electricity purchased from the Ontario grid to supplement hydroelectric power generated at its paper mills. If the system detects electricity price increases beyond preset thresholds, it alerts managers by E-mail and pagers to adjust production capacity.
Demand for real-time information delivery is catching on for some specific applications such as yield-management control in manufacturing and inventory control. There also has been a recent uptick in the use of real-time data analysis, particularly for call-center operations where profiles of customers calling in are built on the fly to assist with cross-selling and up-selling efforts.
Still, the challenge companies face in collecting information from scattered, often incompatible data sources makes real-time business intelligence tough for many. "For a lot of companies, real-time business intelligence is still a few years out," says Tom Burzinski, business-intelligence-practice manager at consulting firm Greenbrier & Russell. Poor-quality data adds to the integration hurdle: Tools for standardizing data and correcting errors before moving it into a data warehouse must work in real time as well.
Real-time business intelligence, in fact, will make IT departments more accountable for the quality of the data they provide for analysis, says Tom Nather, a senior systems analyst at Penske Logistics. "If there are data-integrity issues, they're going to come up much quicker," he says. The Penske Corp. subsidiary, which provides supply-chain-management services to manufacturers and their suppliers, uses a real-time system based on Business Objects SA's Broadcast Agent and middleware from Attunity Inc. to alert suppliers of imminent order pickups.
The question remains whether many companies are equipped to take advantage of real-time business intelligence. "Even if you have these tools, it still comes down to what you can do with the information," says Burzinski. Companies must have the organization and procedures in place to act on real-time alerts and data analysis.
"The bottleneck in the real-time enterprise is almost always human," Teradata's Brobst says. One way around the human factor is to use rules engines to automate responses to real-time business intelligence. A rules engine, for example, can decide how to handle a credit-card transaction that a real-time business-intelligence system has flagged as potentially fraudulent.
Implementing a real-time business-intelligence system isn't easy. Depending on the application, it can involve a broad range of software: business-activity-monitoring tools, business-intelligence software from companies such as Cognos and Business Objects, middleware such as enterprise-application-integration and enterprise-information integration software from IBM and BEA Systems, business-process-management applications from Pegasystems and Microsoft, and data-management tools from Oracle and SAP.
Many vendors are adapting existing products for real-time business intelligence. Ascential Software, Business Objects, and Informatica, for example, have all added real-time capabilities to the data extraction, transformation, and loading tools they sell for moving data into data warehouses.
However, real-time business-intelligence tools that vendors offer are more often for information delivery and alerting rather than real analysis. Analytical-software vendor Spotfire Inc. is working on a version of its DecisionSite analytical software, due out in about six months, which will provide users with real-time analytical tools via a portal, says president Rock Steven Gnatovich.
Another consideration is that real-time business intelligence requires tight integration between operational and analytical software, which is never an easy task. Next year, SAP will try to lessen that burden by offering a new release of its Business Information Warehouse software that can take in real-time data feeds through its SAP XI messaging software.
Some vendors, such as Teradata, are expanding their use of Web services to make it easier to connect data-warehouse systems with live data feeds from operational systems. Celequest, for example, is adding Web services to link its software to enterprise-resource-planning and business-process tools.
To help resolve the human-bottleneck issue, Business Objects is developing collaborative tools, which are due out in three to six months, that will help people cooperatively act on the results of real-time data analysis, says CEO Bernard Liautaud.
Iteration CEO Ken Gardner predicts that as the pace of business accelerates, so will all business- intelligence tools. Says Gardner: "Ten years from now, I believe every business-intelligence vendor will have converted to a real-time architecture. Or they will be out of business."