Industry analysts have predicted that there will be somewhere between 26 and 30 billion Internet of Things (IoT) devices by the year 2020. From personally wearable Fitbit-style activity trackers to electronically empowered fridges, toasters and front door bells… right through to industrial turbine sensors and all manner of corporate monitoring machinery, it’s for sure that we will have a lot of web-connected touchpoints.
More devices logically means more big data captured along the way. But will all that data be valuable data? Is there some way of assessing the inner worth of data so that we can treat it with appropriate worth and gravitas and potentially even trade with it?
Are we capturing some data now (just because we can), but not actually doing anything with it? Should we be recording the entire planet’s every (infinitesimally small) action today, just in case we need to execute some as-yet-uncovered analysis regime upon it tomorrow?
Most of all perhaps, when we do capture huge quantities of big data in a predetermined zone or area, at what point can say that we have reached “enough analytics”, at least for the time being?
The first law of big data usefulness
The degree to which we take the exact depth of big data analytics is directly determined by the corresponding level of insight it produces and where we can still say that we gain ‘productive incremental value’ from doing so.
This is the rarely uttered first law of big data usefulness. This law determines the point of diminishing returns where the analytics effort outweighs the potential gain. But enough theorizing already, this somewhat textbook definition is not that hard to grasp if we look at a practical example.
SAP has made much of the fact that that the World Cup winning German national football (soccer, if you must) team used its ‘Match Insights’ software as part of its training program. Running on the firm’s HANA database management system and software application layer, the German team used this technology to analyse openly available video information captured from cameras around the Brazilian football pitches. Data was crunched relating to player position ‘touch maps’, passing ability, ball retention and even metrics such as ‘aggressive play’.
But did SAP analyse the growth rate of the grass on the pitch, the temperature of the showers in the changing rooms, the cotton thread count in the players’ shirts, the oxygen levels in the stadium corridors and the number of times the manager shouted at the players? The answer is no, it did not.
The point of operational insight
Director of big data at SAP David Jonkers explains that when looking for so-called “operational insights” from the footballer’s data, there is plenty to be going on with just now.
“Once we get to the point where we think we have used all current data to optimize the footballers’ plays and there is no further incremental value we can add, then we might start to look at the length of the grass on the pitch,” he said.
This almost suggests that some data has a latent value, which lies suppressed and dormant until it is unlocked. The company’s SVP & GM for database & technology Irfan Khan has explained how banks do a lot of so-called ‘data hoarding’ due to regulatory forces, but don’t necessary perform a great deal of analytics upon that data as yet.
Data is treated in different classes so that transactional data is generally considered to be high cost. Traditionally, firms would take multiple copies of data before performing any analytics upon it says Khan.
“But in SAP HANA, we work at speed to work on a single copy of data with simplified applications with fewer moving parts,” he stresses.
So it’s all about speed of data movement and speed of big data processing. The opportunity for a footballing metaphor here is too good to pass up. Germany reached the World Cup semi finals in 2010 with an average player possession time of 3.4 seconds. This was reduced to 1.1 seconds in 2014.
German national soccer coach Joachim Löw acknowledges, “Success can never be fully planned; good and bad fortune plays too big a role for that in football. But one can attempt to stem the influence of imponderables as much as possible, with help of good ideas.”
So at what point is big data analytics a waste of time? When not even a German football coach can precision-engineer another goal’s worth of incremental value out of it is the answer.