The Need for Robust Analytics
You might wonder why organizations need data analytics. The answer quite simply is that organizations rely on data analytics to gain a deeper understanding, call it ‘insight’, about their businesses. The understanding could be about their customers, finances, competitors, the market in general, or the employees working for them. Organizations by and large love collecting data. Usually, the larger the organization, the greater is the emphasis on collecting data related to it. It is almost akin to an obsession, albeit of a good kind. An uncle, still alive and healthy, used to collect pebbles as a child; I am told. And all he used to do with those pebbles was that on any given day, he would transfer them from one pouch to another, making no attempt to disguise the loud clack of a count that he would make while transferring one pebble at a time. It’s the same with organizations. If they collect data, and do little with it, but store and just process and present them from time to time in the form of charts and tables, the massive effort spent on collecting large amounts of data serve little purpose. It is not just about efforts wasted, but also about opportunities lost. Data is power; I have often heard this from countless number of people. The same people who look into fanciful charts with the mean and median plodded in and consider the data to be sufficiently analyzed. Well, they could be right if what they are looking for is fairly simple. But it still is a waste of opportunities, of developing a better understanding of the circumstances that an organization finds itself in. I have often wondered why there is a bias and excitement towards the collection of data, but a relative apathy towards performing a rigorous analysis of the data after it has been collected? Is it because data collectors come in cheap, and data analysts (at least the good ones) are a tad bit more expensive?
When faced with analyzing large chunks of data, the usual questions that one wants answered are the what, why and how of things. Organizations often end up seeking an answer to what (mean, median, and standard deviation) and ignore the why and how bit. This is a fallacy, because irrespective of whether the purpose of the data analysis is to conduct a post mortem of an incident or it is to develop an understanding of the business environment, the most important reason why an organization should undertake any analysis is to be able to foretell the future; to be able to predict. This helps an organization plan ahead and be ready.
This is no fortune-telling, as in the process of drawing a rabbit out of a hat. Rather it has more to do with establishing correlations and causations, and sometimes distinguishing between the two. It is about being able to make associations and being able to draw conclusions about the future from them. For example, if an organization manages to establish that the behavior of its managers influences the work/life balance of its employees, and that the work/life balance of its employees is an important factor influencing its employee attrition rate, then it could take measures to influence managerial behavior such that employee attrition rates are impacted in a positive way. This could in turn impact the way an organization goes about its business. And as is evident from this example itself, the very nature of a business environment make it impossible for one to focus on just two variables at a time. For example, it is intuitive that employee attrition rates are affected by factors other than work/life balance issues. It could be that their remuneration influences their decision to stay on with their employer, or it could be that their decision to move was due to inadequate schooling opportunities for their children. That is where multivariate analysis come into play, for observing and analyzing the simultaneous effect of multiple variables on each other.
Image Source – Health and Safety Authority; Rutgers University