A little data here, a little data there? C'mon! Get it together.

by Alan Joch

Vive la différence might be a fine sentiment when it comes to people and ideas, but it can be downright costly for business intelligence systems. The Royal Bank of Canada (RBC) found this out when it maintained data on multiple data marts and an enterprise data warehouse to support its wide ranging financial services business. Although the data marts helped business analysts keep the bank competitive in a changing market, the company decided to take its business intelligence systems to the next level by creating a single view of its data instead of relying on separate data stores. Managing disparate data was starting to feel like "herding cats," says Mohammad Rifaie, senior manager of Information Resource Management and Data Warehousing for the RBC Financial Group, Toronto.

Werner Sülzer, vice president, Teradata's EMEA group.

By centralizing corporate data, the bank could reduce its storage requirements, since many gigabytes of duplicate data resided across the various data marts, Rifaie adds. Consolidated data, in turn, could make the data analysis applications run even faster since programs wouldn't have to churn through any redundant information. The bank also saw opportunities to reduce staffing if it downsized its data marts.

The bank found a way to achieve all of these cost and performance-enhancing goals thanks to a single, enterprise-wide data warehouse built on top of Teradata technology. Now, rather than relying on separate data marts, the company channels its three data collection systems into the single data warehouse. Because data resides in a central location, RBC doesn't have to chase data cats anymore to gather information for its business analysis applications. Centralization also means greater manageability, an end to duplicate data, better privacy control and lower maintenance costs. The bank estimates that the savings in technology and human resources over the next four years will more than pay for what it spent in the consolidation effort. Best of all, the new system doesn't come at the expense of flexibility. Data from the central storehouse can be easily transformed for use by the appropriate line of business as it feeds to "virtual" data marts for end users.

Integrating your intelligence
RBC's story isn't unusual among large, multinational corporations. Data marts often start out as great ideas and become the victims of their own success, according to Werner Sülzer, vice president of Teradata's Europe, Middle East and Africa (EMEA) group. Sülzer sees a widespread movement toward consolidation from customers who had built a data mart because they believed that it was easier and less expensive than committing to a full-blown data warehouse. In time, however, as one data mart turned into dozens of disconnected storehouses, many of these companies—including large corporations like British Airways—began to question the expense and maintenance complexities of keeping the data current and readily available in each of those separate data stores. "You can't have a single version of the truth when you have numerous isolated data marts," Sülzer says, summing up the long-term plight of multiple-mart corporations.
Teradata research shows that 59% of large corporations maintain up to 30 data marts, with some companies finding themselves grappling with as many as 100 separate information repositories. Often, the initial costs for each of these marts are expensed to the individual department that created them. As a result, the corporation has no idea how much it is currently spending on staffing, hardware and software-license duplication.

By contrast, technology research organizations that track business intelligence report that consolidation pays. Gartner estimates that companies that consolidate data marts this year in favor of a central data warehouse can expect cost reductions of at least 50%, while enjoying an increase in business value of 500% by 2004. In addition, Gartner says that through 2006, half of the companies running multiple data marts will spend 10 times more than they would on a centralized warehouse, due to a lack of business intelligence coordination.

Approximately 70% of a data mart's cost goes to data architecture, according to Gartner, with about half of those outlays going for redundant expenses as a company builds new data marts. However, Teradata research shows that by consolidating separate data marts into a central, enterprise-wide data warehouse, corporations can eliminate these redundant expenses by at least $1.5 million to $2 million annually per data mart. In addition, a central data warehouse can achieve an ROI of 400% over three years, a statistic that technology researcher IDC reported in "The Foundations of Wisdom: A study of the financial impact of data warehousing."

Communicating with customers
Across Europe, a number of technical and business factors are coalescing to make the efficiencies of data mart consolidation a compelling solution. Sülzer says the prime consolidation candidates share a handful of common characteristics. First, they're among the world's largest corporations, with revenues exceeding $1 billion a year. Secondly, these companies generate high revenue volume by cultivating large customer bases, typically numbering in the millions. Thirdly, consolidation candidates share a nagging business problem: excess capacity. "Each one of our customers has enough capacity to enable them to generate more revenue with their existing structure," Sülzer says. "The world we're in is defined by huge amounts of excess capacity, which leads to the cannibalization of markets—everybody is trying to poach everyone else's customers."

The resulting challenge for business managers is how to handle the realities of shrinking profit margins and disloyal customers who are likely to switch to a competing product or service based on short-term discounts and other incentives. "These disloyal customers are buying commodity products that have little difference from one company to another," according to Sülzer.

Finally, and perhaps most treacherous of all, these large, data-driven corporations regularly must make multi-million-dollar investments in next-generation gear, such as new aircraft or digital telecommunications infrastructure, to stay competitive in the future. "If you make a mistake in your investment, it's the end of your company," Sülzer warns.

The answer lies in launching the tools to better understand customers' needs for today and tomorrow. "You must be sure that the way you want to go in the future is the way your customers are going," he says. "In the past, large companies hired management consultants to help them make these decisions. Today, they talk to their customers."

Traveling the trends
Talking to customers, of course, is a continuous process that happens formally and informally every business day. The conversation doesn't just take place in highly structured focus groups, either. Corporations collect essential information from all the various business transactions that occur when someone books an airline ticket, orders a new vehicle, arranges for a package delivery, takes out a commercial loan or calls a customer service representative. Each of these encounters provides an important opportunity to better understand customers' needs. "You will discover trends, such as how people use their cell phones, when they use them, how the weather affects usage," Sülzer says. "You look for patterns by region, by gender, by age. You analyze your data and soon you get a profile, an analytic model. And if you study this model over time, you'll see trends that will give you a clear indication of the future."

But when you're working for a large corporation with millions of customers, the sheer volume of information these interactions generate is enough to bury important buying trends and competitive insights under a sea of facts and figures. "You have billions of transactions a day, and they're different day after day," Sülzer says. The key is to store that ever-growing mass of information inside a high-performance data warehouse and apply the tools and business expertise to understand what the data is telling you.

"A data warehouse is something you need in good times and bad," he believes. "Currently, we're not in a boom. Business managers are suffering because their numbers are getting worse. This means they have two options. They can either cut expenses or try to raise the top line. It takes more creativity, talent and energy to raise the top line."

A data warehouse is an essential tool for raising the revenue roof, but the pit of an economic downturn isn't necessarily the best time to ask for new IT expenditures, Sülzer acknowledges. "The boss says 'We're losing money.' But in truth, a data warehouse can help you reduce your expenses and prepare you for the next growth phase. It can help you structure your customers according to which ones really bring value to the business and which ones are causing you to lose money."

Navigating the numbers
Achieving this level of success, however, isn't a plug-and-play exercise. Just as important as deciding to commit to a data warehouse are the subsequent decisions of what technology to use and how to use it. The first step is to reduce IT expenses wherever possible. Often that means either closing down individual data marts or combining elements of them into a central data warehouse. This will save money by reducing physical space, electricity and the number of sites, while DBAs are freed up for other tasks. "All of that is very appealing to the CFO," says Sülzer.

While reducing expenses will also earn points with the CEO, a centralized data warehouse offers an even more appealing benefit for the person in the corner office: coherent data. With centralized information, business managers from the top down get a single view of their customers, even if there are several million of them. This means individual customers aren't counted multiple times in each data mart that might have emerged in the sales, finance, manufacturing and marketing departments. "You need to be on top of your numbers," Sülzer believes. "When you're in recession, it becomes necessary to understand both your historical data from last quarter and also what's happening today and tomorrow. You also must understand your customers. When someone contacts your call center, the service rep needs to know if this is a profitable, or potentially profitable, customer who needs special treatment."

Individual data marts cannot deliver this level of quality information because each mart reflects a different version of corporate reality. "Data marts automatically bring with them redundancy of data," he explains. "Any query you ask of an individual data mart brings inaccuracy, which can mean you'll make the wrong decision based on that bad information." A joint Teradata/BuzzBack .com survey of 108 IT executives revealed that 61% of U.S. businesses operate without a single view of data.

A central data warehouse, on the other hand, combines in-store point-of-sale data with information garnered over the Web, along with supply-chain statistics, into one integrated view.

The second step in achieving success is choosing technology that can handle the demands of a billion-dollar business. Teradata is a fully relational, open-database management system available for UNIX and Windows 2000. To boost processing speed, the recently released V2R4.1 uses a number of statistical functions and complex data mining algorithms that are embedded within the database. This eliminates the need for costly loading and extraction systems necessary for processing data outside the database system. Parallel-processing of all the processing stages in Teradata also allows a large number of computing elements to contribute to the same task to significantly reduce processing times.

In addition, Teradata offers standardized, logical data models for vertical markets, such as financial trading, telecommunications, finance and insurance. For example, the patented financial services logical data model (FS-LDM) consists of eight subject areas, with the customer at the center. Along with the basic module, the FS-LDM also contains special expansion modules for banks, insurance companies and e-businesses. An integrated privacy module provides proper client and data protection.

Feeding the hunger
The result of all this business intelligence horsepower is business decisions driven by accurate data, not gut feelings. "Every business manager is hungry for information because that's the only thing that can guide them to the right decisions," Sülzer says. "If they're trying to make a decision about a huge new investment and they only have data that's months old, how can they hope to predict the future?" But when managers see the quality of data that comes out of a centralized data warehouse, new opportunities immediately present themselves. "It's like a drug," Sülzer says. "The more they use it, the more they want it." T

Alan Joch is a New England-based business and technology writer. He can be reached at ajoch@monad.net.

---

Better, faster decisions in the air

At one time, the airline industry knew that timely departures and friendly service would guarantee success. But those days are long gone, thanks to cutthroat competition and declining demand since September 11. In today's economy, airlines must reduce costs and turn existing customers into loyal customers.

Those are precisely the issues British Airways addressed when it decided in December 2001 to consolidate customer and commercial data onto an enterprise-wide Teradata Warehouse. This wasn't the company's first move to consolidate its information systems. In 1999, it developed the "Ocean Wave" system to replace more than 15 worldwide databases that previously managed customer communications.

As farsighted as Ocean Wave was, it did not provide a single, complete view across the airline of each customer transaction. In order to offer value-added packages and special deals to the appropriate customers, British Airways needed to know more about its clients—when and where they traveled, how much they wanted to spend on airfare, what they needed in terms of destinations, auto rentals and hotel accommodations, and so on.

Perfect platform
After an internal review of its information management infrastructure, the airline realized that analyzing data to improve customer service, increase profitability, etc., would not only be costly but also extremely difficult, if not impossible. That's because information was residing on three different platforms—Teradata, Oracle and IBM.

"We knew we had a large number of warehouses, given a typical airline's investment in IT over 30 years," says Rob Thorne, British Airways' head of customer information. "Our decision to consolidate those warehouses into a single EDW was based on two factors: the opportunity to reduce total cost of ownership for our decision-support data, and the need to make use of a comprehensive, accurate set of books on our business to deliver faster decision-making at all levels."

Teradata wasn't the only option, but it had several advantages. For starters, the airline already used Teradata; by treading on familiar ground, the company minimized the chance of failure and knew in advance it was using a cost-effective solution. Other critical factors included Teradata's scalability, especially since British Airways' has seen its data volume grow at least 50% per year. Teradata Professional Services support team was also a plus.

Last but certainly not least was the solution's reputation within the travel industry. Teradata counts several of the world's top airlines, including Continental and Delta, as customers. Additionally, the Institute of Transport Management highlighted Teradata's industry contributions.

All about the customer
British Airways went high-tech with its platform—a 4-node 5250 Worldmark server running Teradata 4.1 and Teradata's Communication Manager, E-mail Adaptor and Personalization Templates. Although the technology is important, the entire upgrade really centers around the company's most important asset: the customer.

"Our enterprise warehouse will be customer-centric," says Thorne. "We expect the increased level of insights available about our customers and their experience with British Airways to deliver greater relevance in our outbound communication and much-improved levels of service for our customers. And that will help us do the right thing, for the right customer, at the right time."

—Stephen Poole