We are becoming more and more aware that Big Data is just another way of doing “what has been done already” with Business Intelligence systems (BI). So, how does one discern in a practical way if we are really going to benefit from evolving a BI process towards Big Data?
Taking as a reference point the classic rule of the three “Vs” (volume, velocity and variety), underpinning the motivation behind Big Data technology, we are beginning to realise that, in the majority of use cases, neither volume nor velocity changes regarding the processes we have in place. The detection of fraud and segmentation for marketing are two typical cases for using Big Data in insurance. Indeed, they are core areas in today’s insurance business. It can be argued therefore that if we are going to update our processes, using the same data that we already have, then Big Data will not give us any new, relevant results in either of the two cases. Nor is it critical for either case to be performed in real time as opposed to running our routine BI processes overnight.
It will only be worth progressing towards Big Data if we are prepared to take a step further. This extra step involves the integration of new data sources.”
Example 1: drivers’ behaviour
If, when assessing the risk of a driver based on their accident rate and their statistical profile (age, experience, engine capacity), we also analyse their individual behaviour (using a device installed in the vehicle registering the real amount of kilometres driven and driving habits, like in ‘Pay As You Drive’ insurance), this would be adding a new, completely different source of data to the process. We go from assessing the risk with dozens of parameters collected in the quotation form to assessing it using thousands of additional parameters collected in (almost) real time during the period in which the contract is valid. We need Big Data.
Example 2: pricing and retention
A great deal of importance is placed on the analysis of the probability that the client renews using corporate pricing and retention systems. This assessment is limited to dozens of parameters (accident rate of the client, current premium, etc.). Up to now, there has been no way of incorporating what the rest of the market has to offer to the client at the moment in which the client is thinking about renewing their health-care insurance cover. It is something as easy as using a comparison application or quite simply picking up the phone.
With Big Data, we can analyse the competition on a client-to-client level and respond to their search for a better price in real time. To do this, we would need to incorporate into our processes these new data sources, of great volume, and whose format we do not control because it comes from third-party systems. It is in this environment that our current Business Intelligence system can not help us because there are large volumes of data, in real time, from third-party systems, and in many different formats; an ideal ecosystem in which our Big Data can grow and turn it into a success story. Big data use cases are when massive new data sources are integrated.
The butterfly effect in ‘Pay As You Drive’ insurance
We have no idea what will be the detonator that will set off the explosive growth of ‘Pay As You Drive’ insurance in Spain since there are so many parameters involved that it looks unlikely to happen in the near future. In Italy it was a small regulatory incentive that activated the market and in just a few years there are already more than 3 million vehicles with ‘Pay As You Drive’ systems. Virtual SIM cards (IeSIMs) are imminent in Spain and will be key to helping the Internet of Things (IoT) and, above all, Connected Cars enter the market.
Not forgetting that In the United States, in the first quarter of 2016, for the first time ever, the number of new mobile lines for vehicles surpassed new smartphones lines. In the not too distant future a regulatory initiative in Europe will soon require that all vehicles are equipped with an emergency-call system that will, without any doubt, significantly speed up the penetration of the Connected Car into the market.
The result of these initiatives can be compared to the flapping wings of a butterfly; small causes that can have large effects. For example, the exact path taken by a hurricane being influenced by minor perturbations, such as the flapping of the wings of a distant butterfly several weeks later. In this context, the hurricane of ‘Pay As You Drive’ insurance will arrive when the Sales Departments of the Insurance Companies find out what their Actuary Departments already know: if they could collect data on how their insurance holders actually drive this could result in up to a 20 % improvement in the the pooled-odds ratio (profit margin) of the insurance company. Another commercial war on the horizon when the battle for the direct channel has not ended yet.
Over 15 years ago, we witnessed a revolution in the insurance sector: the market wanted to distinguish “good” drivers from “bad” drivers depending on their accident rate history. This paradigm still exists today. A new revolution is imminent: not only can we use accident rates to classify a driver but also we can analyse the way they drive in real time. This will probably be the last change to car insurance as we know it today due to the arrival of self-driving cars.
