SEPTEMBER 2006 (Vol. 39, No. 9) pp. 14-16
0018-9162/06/$31.00 © 2006 IEEE
Published by the IEEE Computer Society
Published by the IEEE Computer Society
Making Business Intelligence More Useful
|BI the BI|
PDFs Require Adobe Acrobat
Business intelligence has long offered the promise of letting companies gather, store, access, and analyze huge amounts of data so that they can make better decisions regarding customers, suppliers, employees, logistics, and infrastructure.
However, BI has had shortcomings that made it unattractive to many companies. For example, the technology's complexity often made it expensive and usable only by technically savvy specialists.
BI often took a long time to yield analyses, making the data unsuitable for real-time usage or short-term projects. In addition, the technology didn't always integrate well with other applications. This made using the data collected by those applications difficult.
Now, however, BI is changing. For example, some applications have interfaces that are simple enough for a broad range of employees to use. In addition, some implementations work quickly enough for real-time purposes.
Additional improvements include vendors using less expensive licensing models and BI being able to integrate into other corporate applications, said Michael Smith, Microsoft's director of marketing for Office Business Applications. In fact, vendors such as IBM, Microsoft, Oracle, and SAP are integrating BI into some of their core products.
The technology is now a hotter commodity than ever. A recent survey of US chief information officers by Gartner, a market research firm, found that companies plan to make BI a top priority in 2006, increasing the budget for the technology an average of 4.8 percent at a time when many other corporate activities' budgets are shrinking or staying the same.
"BI has replaced security as the number one concern," said Gartner analyst Mark Beyer.
As Figure 1 shows, IDC, a market research firm, predicts BI's popularity will increase considerably faster than it has in the past.
Nevertheless, the technology faces numerous ongoing challenges to future success, such as implementation cost and complexity.
BI has its roots in the decision-support technologies first developed in the late 1970s, explained Keith Gile, a market analyst with Forrester Research. Gartner coined and popularized the term "business intelligence" in 1989.
Traditional BI systems consist of a back-end database, a front-end user interface, software that processes the information to produce the business intelligence itself, and a reporting system. The software's capabilities include decision support, online analytical processing, statistical analysis, forecasting, and data mining.
A variety of companies—including retailers, telecommunications providers, travel agencies, and manufacturers—use BI for activities such as customer profiling, customer support, market research and segmentation, product profitability, statistical analysis, and inventory and distribution analysis
The major players in today's industry include BI vendors such as Business Objects, Cognos, Hyperion, and Information Builders as well as major software companies like IBM, Microsoft, Oracle, and SAP.
BI systems frequently consist of multiple elements that don't integrate well together, including best-of-breed components from different vendors. And the systems often work with information sources whose data also doesn't integrate well. The scarcity of widely implemented BI standards exacerbates this problem.
These limitations have caused BI to be slow at gathering and analyzing data, making the technology unsuitable for short-term and day-to-day decision making.
BI products and their interfaces have also been more complex than most applications need and require too much technical sophistication for most employees to set up and use effectively, according to Microsoft's Smith.
"Most of the tools have rich functionality that is only appropriate for about 5 percent of a company's employees," said Forrester's Gile. For example, in financial systems, the functions might be too intricate for all but the most proficient financial analysts.
Vendors also used to offer BI systems only as stand-alone products that did not always integrate well with other corporate software such as customer-relationship management (CRM) and financial applications. This denied companies the opportunity to analyze the valuable information in these applications.
Meanwhile, the data marts required to store the amounts of data necessary for BI operations were too expensive for most companies. For example, a terabyte-sized data mart cost $5 million five years ago, noted Richard Wozniak, IBM's director of BI strategy.
Today, though, Wozniak noted, the use of inexpensive open source software—as well as proprietary software and hardware that are less costly than in the past—has reduced data marts' prices.
Several forces are driving efforts to deal with traditional BI's shortcomings.
For example, making BI accessible to more employees enables a company to spread its benefits to more departments and projects and thus get more use out of the technology investment, noted Michael Corcoran, vice president of strategy at Information Builders.
It also enables employees to perform more effectively, said Gartner's Beyer.
BI systems frequently have been unable to get results to users in a timely manner because of component and data-integration problems.
In large organizations, IT departments have had to gather information from multiple databases—such as those in accounting programs and enterprise-resource-planning applications—and normalize it into a single view in a time-consuming, frequently manual process.
New service-oriented-architecture tools provide interfaces to various data types, which helps integrate data sources so that multiple applications can read them.
iWay Software has developed SOA adapters, interfaces that let multiple programs access numerous data sources. The adapters enable a Google OneBox search appliance to track in real time data stored in more than 85 types of databases and generated by more than 150 transaction types.
With this in mind, vendors Cognos, Information Builders, and SAS are working with Google to use the Google OneBox with their BI systems.
BI's new real-time capabilities can even make it easier for companies to work directly with customers.
A customer might be on the phone or an e-commerce Web site for only a few minutes, which limits the time and amount of information a company has to make sales-related decisions, noted Forrester's Gile. However, new BI approaches can process the information quickly enough to make such decisions.
For example, Travelocity uses BI to analyze customers' input and make hotel, car rental, and other offers to them when they are on the business' Web site or when they visit again in the future, said Mark Hooper, the company's vice president of product development.
BI for the masses
Vendors are making BI easier to work with, principally via interface improvements, so that a broad group of employees can use the technology even for smaller projects.
In some cases, vendors are giving the BI systems themselves better interfaces. In addition, software vendors sometimes incorporate BI tools into applications such as Web browsers and Microsoft Excel because employees already know these interfaces.
Cognos, Information Builders, and SAS are trying to make their BI systems more accessible by integrating them with the Google search bar.
Integrating BI with other applications
A number of software vendors are developing their own BI capabilities and incorporating them into their corporate applications. Other vendors are developing ways to integrate other companies' BI systems into their business applications. These approaches make the data the enterprise applications contain available for analysis and their programs more appealing to customers.
Microsoft plans to include BI capabilities in Office 2007, which customers could access via the software company's PerformancePoint Server 2007, said Smith.
With this approach, Microsoft is trying to offer the entire BI stack: the back-end data mart, data integration tools, front-end BI tools including customized interfaces and hooks into applications such as Excel.
IBM's Data Warehouse Edition 9.1 has added a set of interfaces into its data-management system that make it easier for best-of-breed BI systems to access information stored and managed by IBM data marts, said the company's Wozniak.
Many large companies have terabytes of data stored in their systems from different applications and databases. The average employee cannot use BI to find useful data in this huge volume of material. In response, companies are beginning to use metadata to annotate some of the more useful data to make it easier to find.
Improved search coupled with SOA make it easier to use BI with unstructured-data sources, said Information Builders' Corcoran. SOA parameterizes information so that common data from different formats can be compared. This approach, featured in products such as Information Builders' WebFOCUS BI system, helps map unstructured information into structured formats.
Some vendors place a Web-based interface on top of a BI server. This makes the functionality available to a wide audience via the Internet.
IBM includes Alphablox BI software in its Data Warehouse Edition 9.1. Alphablox lets users connect online to a database, which they can work with via an active Web page that has a Java-based front end, explained IBM's Wozniak.
New licensing model
BI systems' cost has dropped because vendors are using a different licensing model than the per-seat approach they used in the past. Companies used to pay vendors per user, with BI software implemented on each workstation.
Now, though, vendors are using less-expensive concurrent licenses that cover a set number of users. Companies implement the BI software on a server, which users can then access. This approach is less costly than installing software on individual workstations.
BI the BI
Despite improvements, BI still faces significant obstacles to widespread success.
For example, although less than in the past, the cost of deploying a large data warehouse to support a BI system is still high for many organizations.
In addition, calculating BI systems' return on investment is difficult because they provide business-related insights rather than direct links to sales or cost savings, explained AMR Research market analyst John Hagerty. Companies with tight budgets might decide to cut back on such expenditures.
Also, Hagerty said, many companies might resist using something like BI—especially new versions—for the first time because they are unfamiliar with it.
Perhaps the biggest challenge is users' ability to determine how to take action based on the results of BI analysis, said Tom Nolle, president of CIMI Corp., a technology consultancy.
A BI system can present users with organized, analyzed data but they might not be able to make informed decisions based on the information, he explained. For example, knowing that older males buy more of a product doesn't necessarily tell the vendor what it must do to increase sales.
Although BI tools themselves are easier to use, companies still need a technically savvy team to deploy the data warehouse that integrates all their information into one place. The team also must create applications to access the information and decide which data sets within the warehouse will be most useful.
Vendors are adding sophisticated predictive analytics to BI to enable organizations to use their data to better forecast future trends, Information Builders' Corcoran said.
Meanwhile, the BI marketplace will be turbulent as major software companies such as Microsoft enter a field formally dominated by BI vendors, Hagerty predicted.
Over time, said Microsoft's Smith, BI will be more widely adopted as the costs of deployment drop and the ease of use increases.
CIMI's Nolle is less bullish about BI. He said the technology will always entail complex deployment and data preparation and is not easy to link directly to either reducing costs or increasing revenue. He explained, "People expect a tool to produce value on its own. With BI, it may be difficult for that to happen."
George Lawton is a freelance technology writer based in Brisbane, California. Contact him at email@example.com.