Today's business problems are complex and require a higher-level technological sophistication, not only to analyze mountains of data and gather insights but also to look into the future, predict business outcomes, and use these predictions to drive actions. Such an approach to problem-solving is difficult to achieve using simple business intellliegence (BI) tools given the data volumes and complexity.
An inter-disciplinary approach is required, and a process orientation to information analysis involving a combination of BI tools, data collation and collaboration (with external third-party owners of data), integration architectures and technologies and engineering sciences - all integrated seamlessly to handle the problem. This is what we call Collaborative Business Intelligence (CBI).
Business Intelligence, hitherto, has been used primarily to understand what happened, when, where and perhaps, why. Putting together a BI program meant integrating vast amounts of enterprise data from disparate sources using an ETL tool, and generating reports with an industry strength query and reporting tool. These BI tools empowered the analyst by allowing him to slice and dice the data on an ad-hoc basis, thereby helping the business user to get answers to important and critical business questions.
As these questions were often about the past, a typical data warehouse or data mart answered them pretty well, depending, of course, upon how effectively the original data had been modeled.
But in the last few years, many things changed that put paid to the ability of the power user to work with those answers. Some of these changes are:
As a consequence, some important trends have emerged in the business intelligence space. These are:
With data analytics, applied to data organized in data warehouse and data marts, we take the first step towards Collaborative BI. In a limited way, organizations have started applying predictive data analytics techniques to specific aspects of the business to understand what is likely to happen. Popular examples of the use of predictive analytics include: in the telecom industry to understand churn, attrition and fraud; in banking and financial services to predict loan defaults and risk management in general; in insurance and retail to understand consumer behavior. It is also now applied in sports to identify athlete potential and predict win plays; in traffic management in big cities to reduce congestion and promote good road usage; on web sites to suggest articles of interest and in airline reservations as well.
It is fairly obvious that without the data neatly organized and accessible, the statistician cannot provide the insights. However, in many organizations the typical BI analyst and the analytics people do not talk to each other and the functions are more or less separate and not integrated. Going further, within the analytics function, the analyses apply to discrete events and are never combined into a process view. A simple example to illustrate this point is as follows: Multiple attempts to withdraw a large amount of money from an ATM after midnight tend to be viewed by a bank as three discrete events:
The ability to combine these three discrete events into a single fraud attempt is essential for a good security solution and is at the heart of process analytics. While some banks do check back with the customer on their transactions, not all banks combine isolated security events to take a process view-in this case a security threat.