Business Problem Drives Data Initiatives
Data is new oil!
It already seems a cliché now. It is data, data and data all over the place. We often hear people enumerating ‘cool’ accolades in the field of ‘IoT’, ‘AI’, ‘Machine Learning’, ‘Big Data’ etc. etc. and we have new set of professionals, christened as the “Data Scientists”! The sexiest career in 21st century!
Individuals in their daily chore are emanating Terabytes (and growing) of worthy data every day and various organizations are bleeding money through their nose to get hold of that. Across the sector same story, be it healthcare, FMCG, Telecom, mobile and car OEMs, Banks, Insurers, hospitality, airlines, logistics and the list is endless.
Then there comes the techies, the facilitators who help mop up the data from various sources and get them across to whoever wants it. This is predominantly the thriving area of start-ups. There is another key layer as well, and again is mostly start-up driven. These are the analytical firms, the modelers, the pattern decoders, the data gurus! These two sets fearlessly deal with veracity, velocity, volume and variety aspects of any data-set. They actually make the end using organizations, ‘data driven’.
So the importance of data is well established. Hence the justification for efforts and resources that are being put in to fetch and process the same. While, all this was going on, a question stuck me. In the data driven ecosystem, what takes the centre stage? What is the central driving force? Where is the starting point?
We are focusing more on aggregating pile of data and then sweating to see what could be done with the heap of data. While, the data supremacy is well established, have we started putting things in the perspective of ‘Problem statement’? There we go! Even in most of the data driven projects, we often miss to start the discussions with the problem statement. It’ the problem to be solved which should decide what set of data is required, where to get that, how to process that and what to expect in outcome.
While, it’s simplistic, it is often ignored. And the reason behind it is our attitude that unless we complicate anything, the value wouldn’t be optimal. Starting with data and then approaching the solution for a problem is something like, you decide to use a hammer and looking for nailing the solution. Whereas the problem could be solved by digging a hole, which requires different set of tool.
To exemplify, take a case of creating prospect profile for tele-calling in insurance. The sources of data we could leverage, the extent of data that could be gathered from various sources, the profile we could create with the gathered data and how it will help tele-callers in improving their hit ratio. This whole setup somehow is counter intuitive, i.e. to start with data points and end up to the problem. In this context, wouldn’t it make sense if we start with the gaps in current tele-calling; then see if it’s a data problem or a process problem; if data problem, what all data is required to fill the said gap; then the sources where the data would be found; and finally the ways in which raw data would be processed (the extraction, loading and transformation part, ‘ETL’). This approach will surely minimize the effort, time and cost associated with the problem solving. Might require optimal resource base, like would you really require a costly Data Scientist or you could do with a set of data analyst (s) and/or modeler (s).
This ecosystem in question is rapidly evolving and any approach is always debatable. However, the perspective of business problem driving the analytics initiative seems beyond any debate.