In recent years many companies have jumped on the "business intelligence" (BI) bandwagon. What is BI? With access to vast amounts of data, BI is definitely not just a collection of reports generated from historical data. BI is also not software that can generate flashy-looking dashboards that provide readings of current key-performance-indicators but provide no actual intelligence based on extrapolations of past events. The net result is that many companies caught up in the BI hype have been quite disappointed in the return on their investment. To put it bluntly, BI with no predictive analytic capabilites to help organizations achieve high performance capabilities is pretty much useless.
In a series of studies conducted by Accenture, Jeanne Harris proposes a five-step model called DELTA - data, enterprise, leadership, targets and analysts.
DATA - no two companies have exactly the same data even though they may be in the same industry. Mine the data to find what is unique about your company. Use the uniqueness to gain an analytical edge. To succeed in analytics, start identifying the data that you alone possess and recognize the value in it.
ENTERPRISE - data cannot live in silos. Integration of data, processes and analyses in a large, geographically dispersed company is a challenge in itself. So, does it make sense to adopt a companywide perspective for analytics? It depends...especially if all departments in the corporation share markets. According to Harris answers to the following six questions help determine if a companywide approach to analytics will be helpful.
Source: Davenport, Harris, Morison. Analytics at Work: Smarter Decisions, Better Results. Boston. Harvard University Press, 2010
Bottom line - a company that is committed to increasing customer satisfaction will ensure that its data does not reside in silos but is accessible with integrated analytics across the enterprise; the only exception being if data has to be "siloed" for legal reasons.
LEADERSHIP - a company needs analytical leaders - individuals with people skills and an analytical bent. People who know to ask the right questions and will not hesitate to push back when the team makes recommendations based on intuition and not based on hard empirical evidence. Leaders hire smart people, mentor and train them.
TARGETS - Analytics should be targeted at areas where the ROI is the highest and can be easily measured. Some examples are:
analytics that enable effective use of resource sharing;
- analytics that increase customer satisfaction;
- analytics that increase profitability;
- analytics that increase quality;
- analytics that identify dependencies and streamline processes.
ANALYSTS - we need analytical professionals who understand statistical models, probability theory, trend analysis, classification algorithms and simulation. These jobs tend to require an advanced degree in mathematics, statistics and in a quantitative field.
Adopting the DELTA model can go a long way in building a strong analytical capability. However an aggressive implementation of DELTA can also result in stagnation and paralysis if we don't stay away from the following traps. Do not over-fit the curve, says, Kishore Swaminathan, a Beijing-based analyst because there is no sense in analyzing additional data once a pattern has been found. It also does not make sense to wait for data that simply does not exist. Design models and experiments based on the data you have, not based on the data that you would like to have. Lastly, it is important to know the risk tolerance of your company. Do not wait to act unless there is enough data to guarantee a successful outcome.
Business Intelligence when used right will result in an enterprise that is able to make its decisions in a timely fashion based on observations produced by empirical analysis of data.