BI tools enable the data-mining process. They answer questions and provide analysis of abstract concepts that can't be easily implemented through generalized queries, either because of the query's complexity or because the data required to answer the question is derived from multiple sources across the enterprise. Often, for example, financial data must be combined with customer and sales data to achieve the desired results. This means you have to aggregate the disparate data into a centralized location (a data warehouse) or obtain a tool that can access the data directly.
Once the data is accessible, a BI tool provides the means to perform intensive analysis to answer questions across a wide variety of areas, including:
Product analysis: Which of my products yields the most profit? What products are purchased by households with incomes greater than $60,000?
Sales analysis: How are sales trending in the Midwest, where there is a competing store within a 2-mile radius?
Customer analysis: What characterizes the top 10 percent of your most profitable customers? What is the churn rate for customers three months after buying product X?
Query and analysis tools, from companies such as Brio, Business Objects, Cognos, Information Builders, Microsoft, MicroStrategy, SAP and SAS, can provide the basis for achieving important business goals, such as:
Increased revenue: What products might existing customers purchase? What services might entice new buyers?
Increased customer loyalty: What can we do to ensure customers' continued satisfaction?
Increased efficiency: Where can we cut operational costs? How many support personnel are necessary to provide satisfactory service?
Elimination of churn: What are the characteristics of customers who have discontinued service? What is the risk factor of customers with these characteristics? Why are products returned, and by whom?
Not Just for Business
The decision to buy BI solutions is usually business-driven, but perhaps that is simply because IT is unaware of what the tools can do for their own systems and processes.
Most IT departments collect data on how a variety of devices and applications perform using network management systems. However, most of those systems' analysis and reporting features don't lend themselves to abstract analysis. They just tend to report cold, hard facts.
A BI solution could, for example, help IT administrators predict MTBF (mean time between failures) of specific device brands. Using that information, IT staff could buy more reliable devices, and thus reduce the costs associated with failed network devices. Performance, throughput and capacity planning could be analyzed using a BI solution, with the possibility of discovering additional insights previously not conceived of by staff or the network management system vendor.
These tools can also help analyze applications' performance, thus assisting in future platform and development environment choices. Imagine adding a few pieces of data to an existing data set representing historical application performance. Add information about the platform and the programming language in which the application was developed. Now you can use a BI solution to do some extensive analysis that might reveal a poor choice in platform or programming language. A little more research and aggregation of HR records for your IT staff might expose a lack of skills or training in a particular programming language that correlates to a poorly performing set of applications developed in that language.
Such discoveries can help you improve efficiency and reduce the costs associated with downtime. You can determine whether to retrain your development staff, hire more qualified staff or move away from the programming language when a new project arises.
No BI solution vendor claims that its product can make the business decisions for you. The products are designed to correlate data and present it such that people can draw sound conclusions and make well-informed decisions. While some products provide predictive information--the likelihood of a purchase, a response, or a default on a loan, for instance--most do not attempt to interpret the information. That's a business analyst's job, and that's who will most likely use these tools. The DBA and business analyst will need to work closely with the report developers to ensure that the relationships that emerge from the data are used to provide the most valuable information possible.