Evaluation of return on investment (ROI) with SafetyNex
for a Car Insurance Company
 (onboard real time driving risk assessment)

European and American car Insurance Companies are all currently testing onboard telematics systems (on professional electronics devices or on Smartphones), in order to study new opportunities provided by digital technologies in the evolution of their business and business models [1]. We also can see now some experiments in Asia. Indeed, digitization of the economy has an impact on insurance industry too, as new competitors such as GAFAMs (Google, Amazon, Facebook, Apple, Microsoft) come to their car insurance market with new approaches, always ready to capture value. The main idea behind testing telematics is that it would be nice to adjust at best pricing of insurance depending on the driver. Onboard telematics is expected to « measure » usage (kind of road, day/night, number of km, et …) and to estimate risk taken by the driver. Reminder: we have already clearly shown (see chapter REFERENCES) in previous publications that among all these experiments we can tell:

. FAILURE OF THE SEVERE BRAKING HYPOTHESIS

Those systems that only measure driver’s behaviour (accelerations, …) without contextualization (they know HOW the driver drives, but it is unclear WHERE the driver is) are trivial solutions unusable in practice by car insurers, particularly by actuaries who are responsible for pricing. We do know that some solutions developed by insurers themselves (and big ones) are based on this hypothesis of « detecting severe braking » and other behaviour criteria.

We can’t blame them as we made the same hypothesis when we started our research in 2001 ! But believe us, it doesn’t work at all ! and because accident is very rare, i twill take 5 to 7 years to actuaries to detect that they lose money with such telematics solutions ! To be convinced of such a prediction, just ask yourself : consider a brutal driving behaviour (harsh acceleration, severe braking, huge lateral acceleration, etc).

Will it lead to the same driving risk wether the driver drives on a disused airport, in a narrow street in front of a school, approching a crossing road ? … Of course not. And you don’t need to be an expert to notice that, this is common sense. On the other hand, if a driver is very smooth BUT … crosses stop signs without stopping the car, crosses roads without slowing down … then this driver is VERY dangerous. Although this is not a brutal driver with severe brakings ! And you don’t need to be an expert to understand that … again, it is common sense. So please never stop using common sense ! The geographic contextualization of the driving style is absolutely necessary. [2] As a practical demonstration, we used our risk assessment tool SafetyNex and we also used (real experiments with a pilot in a car on open roads) an eco-driving measurement system (EcoGyzer, by the company NOMADIC Solutions) and we compared results : EcoGyzer is very sensitive to driving behaviour. Indeed, when you do severe brakings, then you always need to accelerate after, so your driving behaviour is NOT eco (as you see telematics systems that use severe brakings detection for risk assessment don’t measure a risk … but a kind of Eco driving) . We could see that correlation between EcoGyzer and SafetyNex is ZERO [3]. So, first, common sense tells you that those solutions cannot work, and second, experiments with tools that are already available for sale shows you that it DOES NOT work, definitely.

. FAILURE OF THE GEOGRAPHICALLY CONTEXTUALIZED CLOUD COMPUTING WAY

Those systems offering a contextualization (recording speed and / or acceleration AND geolocation) with the promise of scoring risk in the cloud as a back office service are prohibited by the French Penal Code (Art 226-29) because from geolocation and speed (or acceleration that allows reconstruct the velocity with simple maths compunting), one can go read the points of interest on a digital map (ie. Here, Google, Tom Tom), and read offenses to driving code (infrigements such as speed limit crossing, taking a one way the wrong way, etc) [4]. In France, it is prohibited to record (directly or indirectly) infringements to the law or to regulation codes on computers (except if you are an Administration that is supposed to do that : Police, …). So if you are not based in France you may say this is not a problem. Just be aware of some key points: - European Union is about to launch a regulation pack with the same point of view (more than 20 countries) - France is a 40 Million car drivers country. So if you develop a solution that cannot apply in France, that is too bad, in a market point of view (USA has only 130 Million cars … so you see Europe and even France alone is a big market). - We fully described the SafetyNex solution [5], which took advantage of 15 years of collaborative research with researchers and Road Safety experts, and that now offers a solution proven and unique that: . estimates both usages (time slot, global location by big areas, km/miles, kind of road infrastructure – urban, countryside, highways - , …) , risk profiles, and risk classes.

ALL computation is made INSIDE the smartphone (no indiscreet data is recorded on a distant computer, estimated risk is the « real risk taken by the drive », proven and validated on 50 million km by researchers on road safety community). . uses the cloud only for storing statistics then it is compatible with French law AND with respect for driver’s privacy anyway. . warns the driver in real time before entering a risky area (then the driver has time to slow down), which affects the risk and loss experience : SafetyNex reduces accident rate by 20% (and there is NO other car telematics system that can promise that) [6]. Now, this paper provides an assessment of the return on investment (ROI) for car insurers after massive deployment of SafetyNex.

Read more : http://nexyad.net/Automotive-Transportation/wp-content/uploads/2016/11/Paper-5-En.pdf