Time Value of Data

By | August 30, 2015


“Justice delayed is justice denied,” is a statement we all have heard. Similarly, the concept of the “time value of data” says that the faster the business information is available to the business, the more valuable it is. In fact, the data or information might lose its value completely after certain period. This is not applicable for every situation, but it is for some.


There are many important business examples which come into this category, such as stock price trends, fraud detection, dynamic pricing, cyber security and bio sensors. Here is a common example we all might relate to. Have you noticed lately that credit card companies are able to alert you if your card is not present during the point of sale? Recently, I signed up for a new business credit card from AMEX and as soon as I do any Internet transaction (or over the phone transaction), I get an email with the subject “Account Alert: Card Not Present Transaction.” Similarly, I did an overseas transaction and I got a phone call from a representative within ten minutes of the transaction, asking me to verify if the transaction was something I authorized. How cool is that? Many such business time-value-of-data scenarios can be accomplished by leveraging today’s technology.



The past decade or so, with the availability of data modeling capabilities used especially in MapReduce by Hadoop, new possibilities and opportunities are made possible for business. Through Apache Spark’s open source innovation, it is becoming much easier for organizations to use these technologies and solve real-world problems. In other words, new unexplored frontiers are possible using these innovations. Similarly, SAP’s Hana technology was built in the same manner – through innovations such as in-memory computing. In addition to time processing improvements, what I like about Apache Spark innovations is the attempt to bring Data Science skills to everyone. It is Data Science made possible for the masses. Much complexity in Data Sciences prep-work is abstracted out, and Data Scientists can quickly start doing their real magic.

Trends, Anomalies, Opportunities, Patterns and Risks

Conceptually, using these technologies you can build systems for your business operations, so that you can recognize any of the following:

  • Trends
  • Anomalies
  • Opportunities
  • Risks
  • Patterns

You can easily put in thresholds and tie business actions to the data activity. For example, a retailer can automatically reorder from his or her suppliers based on a reorder point set in advance. Similarly, real time offers can be given to Internet shoppers based on the situation they are in. For example, dynamically giving an instant coupon for a screen protector once the retailer knows that the Internet shopper has bought a smart phone.


While the cost of doing such high impact technology projects is relatively much cheaper compared to the ROI, the cost of not doing such projects is much more. I think the call to action for businesses is to make a list of high impact pain points and start increasing business value through Data Science projects.