Business intelligence solutions are traditionally designed to answer business problems in the structured to semi-structured range. For instance, a financial analysis implementation would try to address budget variance issues, trial balance analysis or income sheet analysis with drill-down capabilities. The solutions are built to take advantage of the higher efficiencies as a result of solid data collection and reporting mechanisms introduced by many technology vendors. A typical data warehousing solution replaces a very inefficient way of reporting in which mainframe standard reports or spreadsheets are used.

One component of decision making that leads to differentiation is the ability to answer complex business problems. Although one may argue that the complexity is in the eye of the beholder, problems such as determining the correct price, where to open the next store, how much to allocate to certain accounts, maintaining the optimum level of inventory throughout the year, estimating competitor behavior, hiring the best candidate for the job or the decision to acquire or merge are categorized as complex decisions. These decisions have a very high strategic impact. What makes these decisions complex is the number of unknown variables that go into the decision function. Decision analysis tackles the problem by introducing assumptions and allocating confidence levels to the variables in the equation. The end result is also a decision that is associated with a probability factor.

Figure 1: Strategic Impact – Integration x Numbers of Users x Complexity

The goal in business intelligence is to provide access to the decision support elements which are derived from the transaction level data collected throughout the organization. Achieving this at one or more process areas and then expanding the scope throughout the organization is one of the factors in the strategic impact equation. The second factor is the number of people with access to the decision support mechanisms throughout the organization. With the first two goals achieved, the organization will search for answers to more complex problems or opportunities. This is when the complex analytics component comes into play.

The traditional model- based decision support can certainly benefit from the concepts of data warehousing. When dealing with complex decisions, it is argued that the lack of data, whether internal or external, adds more complexity to the problem. If the organization had been capturing relevant data for years, why not try to incorporate that into a solution that combines complex analysis, optimization methods and data warehousing concepts?

Figure 2: Two Layers for Complex Analysis

If the solution is geared towards more complicated analysis, it needs a new layer of analysis process added to the traditional data warehousing layer. The reasons for that are:
  • The organization captures additional data relevant to the business process such as product demand and supply.
  • The organization performs certain tests to capture data that is variant in several dimensions.
  • The data that is organized in data models in the data mart needs to be passed to processes capable of model-based and statistical analysis such as regression, decision tree and optimization.

The model-based process application serves many functions such as testing sensitivity limits. The main differentiator of the model-based process application is that the data is now collected in a more proactive way. The internal and external data in the transaction processing systems are built based on the core business of the organization. The data being collected in these applications are passed to the ODS through an extract, transform and load (ETL) process. At that time, we really don’t care about what data we might need, but rather what sense we can make out of the data that we collect from the systems that run our daily business operations. In the complex analytics layer, we actually define the nature of the data we require, but we are not sure about the contents until we get a response from the marketplace. In other words, instead of waiting for the marketplace to tell us a story, we ask a question to the marketplace and request an answer. Customer satisfaction surveys and product surveys are in the first generation of such tools. These surveys however are not performed as frequently as we’d like, therefore, the results can easily get outdated. The true model-based process application constantly collects data and is constantly plugged into the data warehousing layer. With the proactive nature of the new set of data coming into the data warehouse, new decision support elements are to be created.
The second key component of the complex analytics layer is the model-based analysis/statistical analysis component. In a traditional data warehouse solution, the set of reports we get contains detail and summary level data that is derived from the lowest level of granular data that is collected throughout the organization. We then define metrics and key performance indicators, which are calculated from the detail and summary level data. This yields a solid analysis and is linear. The model- based decision support tools and the statistical analysis enable us to run complex models whether linear or non- linear. Conceptually, if the models require additional data, we look for the data sources internally and externally and by means of the model-based process application. The enterprise data model should be updated and populated accordingly.

The overall strategy of the organization should be the driving parameter for design and implementation. For two organizations competing in the same industry, one might want to capture higher market share and the other might be aimed to maximize revenue. For those two organizations, although the majority of the data might be similar, the end result of the implementation must be in sync with the overall strategy. The end result of adding a more complex layer to the existing data warehouse has a direct impact on the business intelligence strategy. The business intelligence initiative should cover all aspects of decision making in an organization, and it should be implemented throughout the organization. The complex analysis that had been done in core business units should now be a part of the overall decision support system. With a better understanding of the complex business problems and opportunities, organizations reach higher levels of decision- making efficiencies, hence maximize the strategic impact of the business intelligence solution.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access