Sifting a Riverbed of Data for Insight
Pizza-flavored cheddar cheesegood business opportunity? Yes, as Cabot Creamery determined. Does osteoporosis more often hit wine drinkers or teetotalers? Imbibers, as Kaiser Permanente can tell you. Will women who buy lace dresses by mail also buy lace bras that way? Not really, as Victoria's Secret knows.
Business intelligence software revealed the answers to these questions by synthesizing data that already existed at these companies, sometimes mixing it with outside demographic information, then analyzing it six ways to Sunday. The companies then made business decisions based on the results. Cabot started selling a new cheese, Kaiser looks for bone problems in heavy drinkers and Victoria's Secret knows not to push brassieres on certain customers.
Yet it isn't easy to sort through the hundreds of business intelligence products for sale. SAS Institute, the biggest vendor in this market and a specialist in statistical analysis tools, itself sells more than 50 products.
Some key categories include: tools to query back-end databases and build reports based on the results; data mining tools to pull specific variables from data sets; and online analytical processing (OLAP) tools, which are statistics tools that rely on multidimensional databases. Even everyday spreadsheets can do a lot of this work. Bankers, especially, love the spreadsheet method.
Product lines of the major vendors overlap in some areas and stand alone in others. Business Objects, for example, is known for easy-to-use query tools. Cognos has more complicated query tools, plus OLAP. Hyperio offers both an OLAP database and analysis applications designed specifically for finance tasks. Information Builders has a report writer and tools for gathering data from different sources.
Different tools are needed, depending on the kind of data available and what you want to find out.
For example, analysts would use OLAP to discover that there is a general tendency for two or more things to happen together. Beer, for instance, might sell better on hot summer days to men in their 20s. Identifying the population of those people in a geographic area to initiate a marketing campaign calls for mining of a demographics database.
Kaiser Permanente and other health care and pharmaceutical firms use statistics tools from SAS and various OLAP products. They study complex drug trials and patient medical information, involving dozens of variables that require traditional statistics techniques. As Sue Emmons, director of data intelligence projects at the Vancouver Coastal Health Authority, puts it, "We're not into widget management."
But other companies are, in essence. CVS Pharmacies, for example, tracks product sales at its 4,200 stores with an application built on Hyperion's OLAP database. Six hundred managers use it to find trends and streamline inventory.
A Shifting Landscape
Big applications vendors are moving into the business intelligence space, upsetting the traditional per-user pricing model that business intelligence vendors like.
PeopleSoft, Siebel Systems and SAP, among others, now have basic analysis features in their enterprise suites; IBM, Microsoft and Oracle have added or plan to add analysis capabilities to their databases. All of these applications are typically sold via enterprise licenses. Now that some form of analysis is included in those flat rates, some users question whether they need separate, expensive tools whose licenses they must manage and track, user by user.
Alta Resources, a call-center outsourcer in Neenah, Wis., has chosen PeopleSoft's soon-to-debut analysis features over third-party tools. It's better to have such capabilities as part of Alta's core enterprise applications than to have to sweat the integration of two vendors' products, says Bill Parry, chief technology officer at Alta.
The response of tool vendors to the challenge is twofold. First, in true tech-industry style, they are partnering with those that threaten them. Tools vendor MicroStrategy, for example, has a deal to bundle limited versions of its analysis products with enterprise applications from Retek. Cognos has a similar arrangement with PeopleSoft.
Second, several tools vendors now offer analysis suites for specific vertical industriestelecom, pharmaceuticals, bankingand horizontal tasks like customer service, supply chain management, and procurement.
"Instead of the IT department buying a bunch of tools and building their own apps on top of them, we're seeing a trend toward packaged analytic applications," says Dan Vesset, an analyst at International Data Corp. "The theory is, you buy it and it runs."
Technology managers must carefully consider a vendor's financial health along with its technology savvy.
"Stability matters a lot right now," says Gartner Inc. analyst Howard Dresner. "The old economy is back, and it's pissed," he says, referring to the fact that many tech vendors are reporting financial losses or are going under.
Financial troubles at MicroStrategy knocked it out of the running at Sara Lee Bakery Group, says Chief Information Officer Steve Brazile. In 2000, the bakery was looking for an analysis tool for 500 users to formulate pricing and promotions. At that time, MicroStrategy had slashed staff by 60% and lost $261 million for the year. "We didn't feel comfortable that they would be around," Brazile says. He went with Business Objects instead.
Another reason to choose carefully: Once a business group adopts a tool, it's darn tough to replace it. Analytical tools are quite complex. People build whole careers around specialized skills. "It's akin to telling a Cobol coder, 'You're going to write in Java now,'" says Brian Kilcourse, chief information officer at Longs Drug Stores.
By no means do you need business intelligence tools to do business intelligence, says Ellis Moore, vice president of information technology at Nobel Biocare USA, a $250 million maker of dental implants. Moore evaluated products for eight months before deciding none would do. The company wanted something that could easily extract data from SAP enterprise resource planning systems, relay it to a relational database, then layer it so sales agents could get more detailed information via Web browsers and a dial-up connection. Nobel Biocare built the system itself.
"A lot of people just buy the tool based on a vendor demo," Moore says. But don't fall for demos. His advice: Grill salesmen and women on whether the products can manage your particular data sources.