Make Data Actionable for Decisions
Keith Gile of Forrester Research talks about the functionality of business intelligence.
Oracle Magazine spoke with Keith Gile, a principal analyst covering business intelligence (BI) at Forrester Research, about new strategies for corporate decision-making.
Oracle Magazine: How do you define BI?
Gile: BI is the process of making data actionable for the purpose of making better, more timely business decisions. It includes technologies such as online analytical processing (OLAP), data mining, ad hoc query, and enterprise reporting, but these terms are germane only to the people who produce BI applicationsa small community of users that represents perhaps 7 to 10 percent of all possible end users. Today's BI applications are reaching more types of business and casual usersdata and information consumers. These decision-makers want to run reports and visualize data, but they don't care if they have an OLAP or a data mining tool. They simply want to do their jobs. Often BI technology ends up under the covers as part of a dashboard or some other type of application that's been created to facilitate the tasks of business users.
Oracle Magazine: How do you categorize different BI applications?
Gile: You can divide BI applications into strategic, tactical, and operational. Strategic BI is analysis after the fact: You capture transactions from applications and then model and aggregate that data for reporting. Tactical BI pushes decision-making closer to the cusp of when a process is going to happen. Operational BI allows decision-making to occur within a process and can even trigger a process. Operational BI applications analyze real-time or near-real-time data via an operational dashboard or by being embedded directly in an operational workflow.
Oracle Magazine: What are examples of each of these types of BI applications?
Gile: At a strategic level, a BI application might help the VP of marketing understand the implications of a promotion and how it will affect revenue over the next six months. In support of this promotion at a tactical level, a call center manager might use a BI dashboard to help call center reps stay abreast of day-to-day changes in the supply chain while trying to manage the performance of all call center reps. At an operational level, the individual call center reps might obtain information about customer orders and product availability to determine which offers to pitch to a customer from one moment to the next.
Oracle Magazine: What are the pros and cons of embedding data mining and other kinds of BI functionality within the database itself?
Gile: The advantages of embedding OLAP and data mining within the database revolve around the administration and optimization of those data mechanisms. You no longer need to store data in a separate proprietary analytic engine; you can store detail-level data in a relational environment, with a direct link from the multidimensional cubes to the relational tables. You don't have to move as much data or maintain redundant data stores, and you can optimize one database rather than coordinating multiple databases, so fewer people are needed to perform database administration and data modeling. On the flip side, embedded BI tools may lack the advanced functionality that you find in a pure OLAP or data mining engine. For example, when data mining is embedded within a relational database, you might have only a limited set of algorithms available or a limited capacity for descriptive and predictive types of analysis.
Oracle Magazine: Customers want to analyze both structured and unstructured data. How are vendors responding?
Gile: Data can be analyzed only when structure is applied to it. That's why we're starting to see mechanisms that let you declare structure around content. For example, segmentation tools can look through documents to discern keywords and other important features, create a structured taxonomy, and then join this information to structured data in a database so it can be retrieved and analyzed with a standard query tool. Most pure-play BI vendors lack content management capabilities, and most content management vendors have no BI capabilities. We need more-flexible analytic mechanisms, and we need data repositories that can extend the reach of data in many different ways so that all data has similar value to end users.
Oracle Magazine: Are we seeing a consolidation in the number of BI tools that companies want to use?
Gile: Our clients are saying, "We have too many tools and not enough solutions." In response, vendors are developing BI platforms that include both analytic and enterprise reporting, OLAP, data mining, portal integration, and visualization capabilities. Most companies don't want another reporting tool. They want an enterprise solution that transcends functional areas and integrates these tools with facilities for application development, database integration, and deployment through portals and dashboardsideally using service-oriented architecture.
David Baum ( email@example.com ) is a freelance business writer based in Santa Barbara, California.
Forrester Research ( www.forrester.com ) is an independent technology and market research company that provides advice about technology's impact on business and consumers.