With the buzz around business analytics, its not surprising that companies struggle to find value with all the data they collected. Analytics is much more than just collecting data and writing reports. The concept of business intelligence is around collecting data and understanding the existing environment, but analytics requires companies to go much further, and begin to provide accurate and meaningful predictions of future behavior and explanations of present activities.
There are two primary components to a strong analytics program. I call this the Yin-Yang Analytics(tm). This approach refers to the never-ending cycle of intuition and mathematics as the cornerstone for a good analytics program. It’s not about the technology, but what you do with it that makes it count. Intuition provides a foundation for where to look in the data. However, within intuition, a number of biases exist and expert experiences tend to cloud new insights. The other half of the circle, mathematics, is not only informed by the expert based on the intuition, but provides insight to the expert that confirms previous suspicions or identifies areas not under consideration previously. This cycle continues as the mathematics feeds intuition and intuition feeds the mathematical process.
Effective analytics takes into account the variability of the data, and its’ multidimensional aspect.
Effective analytics takes into account the variability of the data, and its’ multidimensional aspect.If you are looking only to predict single variables, you may be missing a bigger picture. Furthermore, many analysts just look at correlations between variables without exploring the intricate and complex relationships of these individual data points.
Some of the questions you should be asking you analysts don’t need to be “statistical” in nature, but intuitively one might ask whether the analysis is focused on a single response such as “sales”, or whether the analysis looks at sales and profit margin simultaneously. Many tests in medicine and other areas look at these dependent combinations rather than just single outcomes. Business is not a single dimensional, or single point construct. It is a complex combination of inputs and outputs.
Some companies focus on visualizations as the key to analytics. They produce nice charts and graphs for departmental heads. Having nice visualizations is always a plus, but analysts should be very careful not to hide the truth in the form of charts. Anecdotally, analysts consistently hide information from their superiors and bury results so as not to arouse suspicions, a very dangerous proposition. In the absence of a clear understanding, by executive management, of how the analysis was conducted, how it was derived, and, more importantly, how it was validated, obscuring results will become more prevalent.
In short, you should understand what your analytics team does, how they conduct their tests and analysis, and how they validate results. Your analytics team’s results should easily translate into actionable outcomes and not just be mathematical exercises. Best results are achieved using a cyclical dynamic process with subject matter experts and analytics experts working collaboratively experts toward the common goal of searching for truth in data.