“Stretching” refers to the act of deploying data for more than it is usually used for
This is the fifth in our series examining Data and its idiosyncrasies. In the last four instalments, we looked at Data Visualization and Stories, Thick Data, Data Confluence, and the Data-Discovery-Disruption cycle. In this viewpoint, we will look at the opportunity existing Data can provide if we stretch it (and our minds) a little. “Stretching” here refers to the act of deploying data for more than it is usually used for (and not to the Azure product or to a technical function in some of the visual processing software). Some of the concepts and examples here are parts of industry in the form of Data Acceleration, Data Extension, Data Mining, etc.; this column simplifies all of that into a single stream of thought with some simple examples.
Data is more than it looks like. There is more in it than one has collected and ingested
We will not get into any of the techniques mentioned above but will look at it from a singular and logical point of view. And some of our examples, while falling in the realm of these techniques did not actually use any of those techniques. All of these were based on simple business logic. Data can be stretched in one of the four ways: Vertical, Horizontal, n-th Dimension, and Across. The contention here is that any dataset can use all four of these methods; together or separately.
A couple of things should be noted here first. The benefits of stretching the data can only be realized by developing new capabilities, in product development, service provisioning or supply chain management. Secondly, some of the examples below happened by intuition or accident. But today rigid self-proving and self-learning Data Science models (that use statistics and calculus) are being deployed to exploit these opportunities.
One way to stretch the data is Vertically. One goes further down the same pattern or the path the data is “obviously” showing, and project based on that. As an example, a very large retailer in India discovered through data that milk and sugar are usually purchased together. And intuitively, the retailer’s IT department looked at where this was not true. There were about 10% more customers who were only buying milk. A focused campaign on this subset of customers brought them back into the stores to buy sugar also.
Second way is to stretch it Horizontally. Use the data to expand the business into other business areas. Banks had the data that gave them a fair idea of the income and net worth of their customers. A simple analysis provided them with a list of customers who are likely to be investing in stock markets, insurance, etc. And they quickly moved into these areas. Today, every bank in the world offers investment and insurance brokerage in addition to a host of other services.
A third type of data stretching would be to take the business into n-th Dimension. This has come into focus especially during COVID. People stuck at home with lockdowns switched to video/online games as an outlet. And suddenly data showed that these were being used by 30-50 year olds in addition to 20-30 year olds. This new group obviously has the highest purchasing power of all the age groups. And so, the games industry has now added all kinds of services in its advertising: financial services, retail, FMCG manufacturers, and what not. And they did not stop there, they incorporated e-sports, a perfect replacement for the sports games that this consumer segment was missing. This increased the volume of betting inside the games by a very large multiple. So, now data showed that the demography of its user base was shifting, and the whole industry changed. And the rest of the players in the economy have now started pouring money in advertising on these platforms.
Lastly, there is the stretching the data Across demography, geography, product, situation, etc. You take the learnings of data from current scenario and replicate it across a different demography, geography, product, or situation. The belief here is that people are similar everywhere and related products behave similarly. As an example, one of the largest retailers in the world had perfected the art of rejigging its supply chain to cater to changed needs after a hurricane hit an area. Data taught them what is needed most in a disaster: water, bread, eggs, shoes, underwear, etc. So, when COVID hit badly in certain parts of the US, this retailer was the only one who had its shelves restocked with what the consumers really needed (toilet paper notwithstanding).
Bottomline is, Data is more than it looks like. There is more in it than one has collected and ingested. IT departments need to ask the following of their data:
- How far can I push what the data is already telling me?
- Can this data be used beyond its current source?
- What is “new” in the data?
- Where else can I generate the same data?
The author managed large IT organizations for global players like MasterCard and Reliance, as well as lean IT organizations for startups, with experience in financial and retail technologies