Data Strategy in times of extremes!

As most of us are adjusting to this new normal of “shelter-in-place” and “working-from-home” with the fear of an invisible intruder who can enter our bodies and make us sick-or-worse, there is a significant drive by all sectors of our society to bring things back to the yestermonth’s state of affair. From day one on this Pandemic one has become painfully clear... critical data that is required to make intelligent decisions is either not available or incomplete or too late or prone to manipulation. In addition to the data issue, we are also challenged with the ability to churn the data, run models on it, and draw good intelligence out of it. While some of these issues are unsolvable, there are certain things that we can learn from our collective experience in making sense of how the data is being used in making decisions and how some of these decisions may or may not be correct due to lack of good data, processes, and governance of information.

Most analysts believe that there will be lasting impact of Covid-19 pandemic on society at large. I do hope that we do learn from this experience in giving the adequate importance to one of our most important enterprise asset - DATA and build a sustainable strategy that can transcend normal and extreme times. At its core a good data strategy is like good plumbing or electric wiring that your building has... you only know the importance of it when things are not work; but you better not ignore it. Listed below are some of the key lessons that we should be learning as we are going through our current transition in the way of our collective living.

• Data is truly critical for an enterprise and it become more importance in the hour of extremes
• A model is only as good as the underlying data. You can have the best model in the world but if underlying data is not good... model outcome will not be trustworthy
• Data sourcing strategy and data supply chain is as critical as product sourcing and product supply chains
• Keep adequate history of your data and model outcomes; it is possible that current challenge that you are facing has similarities with a situation that is not from recent years or even decades
• Security and privacy is critical for data however situations may drive companies to more data sharing and less privacy
• Domain knowledge and Human intelligence is utmost important in extreme situations and understanding outliers
• Lastly, investments in data practice during normal times will result in positive outcomes when companies are in extreme situations

Broadly following diagram shows key components of a good data strategy and practice. It is important to understand that AI/ML model creation and model area management are and should be part of overall Data Consumption layer where companies. 

Hopefully we are going to come out stronger and resilient and above all more learned and prepared for future.