The main value that analytic applications bring is built-in business intelligence, which is a predefined set of metrics and key performance indicators that business users would have had to build otherwise. By purchasing an analytic application, organizations also purchase business expertise with which they can gain valuable insight into their businesses. (See Fig.1 Processes and applications.) True analytic applications contain the...
key components of business intelligence systems such as ETL, standard and ad-hoc queries, alerts, etc. The best applications offer three additional functionality elements:
- Complex analysis
- Decision deployment
- Closed-loop analysis
The complex analysis component allows for detailed statistical analysis of the data collected in the data mart for solving a specific problem such as pricing and inventory problems. The results are collected in the data mart along with the internal and external data, and are then presented to the decision-makers in forms of conventional visualization methods. The decision-maker submits recommendations to the application system, and then the application submits a decision to the operational application systems, where the decision is carried out. The analytic application collects data from the operational application continually, at some point picking up data relevant to a decision or suggestion presented to the operational application as mentioned earlier. At this point, the loop is closed. The analytic application performs the analysis not only based on internal and external historical data, but also based on data that contains the implications of previous analysis. In the best analytic applications, the technology and the data framework is built such that when the decisions are carried out on the business side, the application has the capability to tie that decision and the results back into the historical data which was the source of the decision in the first place.
(See Fig. 2 Analytic Application Interactions.)
One challenge with analytic applications is that they are very focused in terms of the business processes for which they provide decision support. A CRM analytic application will not be helpful in questions about finance or manufacturing. However, the nature of decision-making process is such that two things are likely to happen: the decision-maker will need data from more than one business process, or, after initial analysis, the decision-maker will increase the scope of analysis trying to find solutions to more complex decision issues. When either of these scenarios occur, one of these three things must take place:
- The decision-maker manually collects and analyzes data
- The decision-maker tries to use a different system
- The analytic application automatically extends its scope
The second challenge of analytic applications is integration with the heterogeneous source systems. The claim of analytic applications is that they are more cost effective than data warehousing systems. This claim assumes that data issues such as integration, cleansing, etc. are not the main responsibility of the analytic application. Data issues are the main reason for failure in analytic projects. So one end of the spectrum is several applications from several vendors for different business and decision processes, and the other end of the spectrum is one single application for all. It is crucial for a given organization to decide where to position themselves in this spectrum of applications. This decision is based on the human and capital resources and the urgency for the application. It's key that every organization have a framework for information technology and decision-making. After that, the acquisitions for technology and knowledge must fit or be able to be adapted to the existing framework. Black box applications do not fit in this criteria and the total cost of ownership may be very high. This doesn't mean that all the components of the application must be shared, but the data collection mechanisms and data storage must be open and must allow for integration with existing data marts.
In summary, with all the pros and cons, analytic applications offer value to organizations with decision issues. They differ from business applications in that the level of involvement during implementation is far beyond typical IT involvement. Although analytic applications come with a set of predefined metrics, some custom metrics might still have to be defined and the exercise of identifying key performance indicators from the long list of metrics must be carried out very carefully and thoroughly. Integration of the analytic application with the overall business intelligence framework is another key factor in securing a positive return on investment. To this end, the analytic application must be "open." Ultimately, the main reason anyone needs a business intelligence solution is to ensure competitive advantage. And for a fraction of a major scale business intelligence implementation cost, analytic applications provide insight and decision support capabilities given that the core functionalities mentioned previously exist in the application.
Mehmet T. Oguz is the architect for the Pricing Data Mart and Analytics at Zilliant, an Austin, Texas-based pricing and revenue management company. Prior to Zilliant, Oguz designed and implemented business intelligence solutions for several Fortune 500 organizations as a consultant and systems integrator. He can be reached at firstname.lastname@example.org.
Fig.1 Processes and applications
Fig. 2 Analytic Application Interactions