What is Business Intelligence (BI)?
Business intelligence. Is that an oxymoron like jumbo shrimp or clean coal? It is more commonplace that you might expect. Business intelligence (BI) is a set of software applications used by managers and analysts to explore their internal data to support business decision making, whether that is seeking business opportunities, internal inefficiencies, or operational problems. It covers the range of systems from the simple to the complex, from downloaded data in Excel that creates a summary report for one manager, to a server farm running specialized ad hoc query and visualization software that support thousands of store managers for Starbucks.
Once a company's data is organized, analyzed, and presented in a useful format, managers can make intelligent decisions faster. This includes decisions as diverse as allocating heavy equipment among construction sites, scheduling staff for a call center, and opening a new distribution center. By using data that already exists inside an organization, BI systems can lead to more effective use of scarce resources, reduced expenses and improved revenue. It just makes sense to take advantage of information that you already have on-hand to make better decisions.
Business intelligence is big business: approximately $14 billion in 2012 (Information Management, 2012). Vendors want to sell tools or consulting that provide parts or all of this: OLAP from IBM and Microsoft, dashboard builders from iDashboards and Jinfonet, build-it-yourself tools from SAS, high-powered hardware/software for big data projects from Oracle, project management consulting from Accenture. Managers and other end users may see BI as a dashboard or ad hoc query tool.
7 Things to Know
1. BI is nothing new. It is just IT creating a way to support managerial decision making. That has been happening for many years, starting with the first summary report.
2. The format of the user interface can vary tremendously:
• Summary reports and exception reports
• Overview with drill-down capability in numeric and textual interfaces
• Graphical dashboards with visual information
• Data Visualization, e.g. maps and multidimensional representations
• Real-time access via mobile systems
• Ad hoc query, even from an Excel spreadsheet
3. You know you need BI when your spreadsheet models are everywhere, and show different results. Excel is a flexible data manipulation and presentation tool, but lacks sufficient power for large data sets.
4. Keep BI off the transaction processing system for best performance, although it may exist there briefly in order to gain access to real-time operational data.
5. Consider your development team's skills when choosing tools for building your BI system. Having a hot shot team of ASP .NET programmers who build great user interfaces means you can leverage their expertise. A team of COBOL programmers will not be so helpful, though they can probably extract and transform your transaction data.
6. Choose components that are widely supported. If you build your data warehouse in Oracle or SQLServer or even MySQL, you will have no trouble finding user interface components – ad hoc query and web-based dashboards – that will work well.
7. The time horizon for a new BI project will depend on developing three things: clean data, familiar tools, and a great user interface. If these three components are built separately, you may have trouble as they need to work together. We once worked on a large logistics system in which management decided to delay building the database from legacy data fields. When they finally decided to build the database, it was a much bigger project than they expected. This caused the user interface and tools teams to wait with nothing to do until the data was ready.
We will continue our discussion of business intelligence with a case study on designing a dashboard-based system for a big bank. Until then, have fun!