February 20, 2017 | Written by: Karen Lewis
Share this post:
At the Genius of Things Summit, Guy Raz, a National Public Radio (NPR) moderator, hosted a panel discussion between Rob Bauer, Managing Director, Strategy & Innovation, AIG, and David Knight, Founder & CEO, TERBINE. The panel’s focus was to discuss monetizing data and standards.
Sensor-driven pricing is the future of insurance
Rob Bauer from AIG was first to comment that the move to sensorization is poised to transform every industry in the world, not just insurance, which is why organisations are racing to figure out how to monetize IoT data.
Insurance is one of the pieces of oil – like data – that helps coax the economy along. For the insurance industry, the move to sensorization is a big jump from where it used to be. To illustrate the point, Rob asked the audience for a show of hands on how many audience members had owned a red car? Not surprisingly, a significant proportion of the audience said yes.
Well this was bad news for anyone who raised their hands because in the old way of doing things, if you owned a red car the insurance industry would have considered you a high risk driver – deemed a worse risk driver, prone to aggressive driving, simply based on your colour preference, plus answers to a few other questions. But is this decision actually true, auditable or predictive?
Can data change behavior?
In the insurance industry, having pricing certainty is Nirvana. In the era of sensorization, it’s now possible to make more accurate and quantifiable assessments using real time usage-based information. But, how do organisations like AIG evaluate how real-time, sensor-based data can be used as intelligence?
Putting sensors into objects results in personification because the sensors give feeling to inanimate objects. AIG believes the same principles can be applied as a framework using:
- sensorization (the sensors);
- thinking and analysis of the data;
- stimulus response (application of brakes to stop, giving data to someone from a smart watch, a machine telling the operator that it is going to break down before it actually does).
Testing a hypothesis
Sticking with the automobile industry as an example, we can see how a vehicle contains thousands of sensors which actively collect data. To test a hypothesis, AIG worked with one of its clients, Europe Car, to launch a smooth driver test. In January of last year, anyone renting a Europe Car in Ireland was offered an opportunity to participate in a Smooth Driver Pilot Program.
AIG and Europe Car measured smooth driver behavior using four criteria:
(1) how hard a driver brakes,
(2) how the driver corners,
(3) if the driver is prone to speed,
(4) how fast the driver accelerates.
During the 17 week pilot program, rental cars were installed with sensors to collect data to measure ‘smooth’ driving habits. The data collected was sent to the IBM Cloud where Watson IoT analysed it in order to calculate an individual driver’s smooth driving scores. Once given a safe driving score, participants of the program were encouraged to play a game where they competed against each other to become the safest driver of the week. The winners were then entered into a competition to win an Apple iWatch.
Result: 23% reduction in claims
Guess what? Over the course of the 17 week pilot program, AIG experienced a 23% reduction in claims. That’s a result. From this exercise, Europe Car and AIG proved their hypothesis – data can be used to change behavior. But, does this mean organisations using IoT data could develop more accurate pricing models based on a combination of traditional and contextual data sources? AIG thinks the answer is yes.
Following data breadcrumbs to help clients save money
Many insurance companies can use historical claims analysis to help clients reduce claims. Exploring claims runs – or spikes in claims allows organisations to follow the data breadcrumbs to understand the source of the run.
In one example, AIG examined a spike in claims submitted by one of their hotel clients. After investigating the run of claims, AIG discovered the housekeeping staff had been submitting claims after sustaining injuries while making beds in rooms where the night tables were attached to the beds.
Based on the intelligence gleaned from the claims run, recommendations were made to the client to change the working environment. By detaching the night tables from the beds, workers experienced fewer injuries, resulting in reduced claims. In this instance, a simple change meant fewer work-related injuries, and a smarter system that mitigated risk for all concerned – the insurance company, the client (hotel), and the employees.
Creating a commodities exchange for data
If data is the new oil, then TERBINE believes the world is ready for a new kind of commodities exchange – just like the financial world where oil is traded – only for data.
Taking into account things like provenance, context, and known quality of the sensors, not to mention data ownership and regulation, TERBINE is positioning itself as the commodity exchange for IoT data. To help realize that vision, TERBINE is taking advantage of IBM Cloud, Watson IoT and AI to facilitate a means to monetize the massive volumes IoT data available.
Making sense of tertiary data
Sensor data is time series data stored in huge CSV files. It could include anything – geolocation data, surface water temperature data from the Indian Ocean, a count of cycles moving through the congestion area of London – you name it.
Silo or rubbish – it’s a new twist on recycling. Rather than throwing away tertiary data, organizations like TERBINE want that data – to make sense of it. A car for example is a perfect vessel to collect tertiary data – cars are literally a rolling sensor platform. By analyzing data collected in vehicle sensors, we can learn things like where drivers accelerated, where they swerved, where a car hit a bump. The aggregation of this information, when combined with geolocation and other data points can then be used as intelligence for other stakeholders or participants in the overall transportation ecosystem. For example, it could be used to inform road maintenance services of potential driving hazards like potholes.
Finding the signal in the noise
With Watson IoT Platform, TERBINE now has the infrastructure to collect tertiary data from all over the world. To handle the zettabytes of tertiary data being created, TERBINE created a meta data construct which they intend to make available to the industry through open source, which can then be used to help organisations find the signal in the noise.
Context is the difference between a data point and intelligence. If you know you are traveling at 40 miles per hour, that’s a data point. But, if the speed limit is 20 mph, you’re nearing a school area, and it’s snowing – that’s data in context which can be used to inform next actions. By adding things together, connecting the dots, and then taking things away – you can test things – making the use of data much more of a science.
Charging the right price for the right risk
By adding context to data points, organizations are in a better position to charge the right price for the right risk. To price any data, there are 20 variables that impact it – only one of which is timeliness. Using AI, the data can be dynamically price based. But, should behavior drive pricing? If you analyse risk in the rear view mirror, you can’t; but, if you can take information in context – looking at the sum of its parts – it’s possible.
In the age of sensorization and data monetization, we are creating an environment that incentivizes manufacturers to introduce new technology into end products – things like sensors in cars, heating systems, clothing, etc. But to what purpose? After the data from all these sources is collected, analyzed in real-time and aggregated, it is used to inform an entire ecosystem in a way that enables more accurate pricing. But, does trading data on an open market open up all sorts of other questions about data privacy and data cross-use?
The headwinds of privacy, regulation, and ability
From David Knight’s perspective, TERBINE scrubs out any of the personalization from the data they are collecting. By the time the data is aggregated, anything personal is removed. The issue of privacy gets more complex with the cross-use of data. Cross-use data refers to that which is being used by organisations who don’t own that data – but who would benefit from using it. An example might be an automobile, where data has been collected using devices installed in competitive vehicles – air bags, for example. Once data makes it to the commodities exchange, how is it possible to track it from its provenance to prevent competitors from gaining accessing each other’s data? On this point, both AIG and TERBINE agree that blockchain could offer a flexible solution capable of tracking who receives the data on the other side, while managing who is able to see or receive that data depending on who or what they are – for example, a competitor, or a regulatory body.
Digital trust is imperative
As consumers, we are now willing to accept that when we want a taxi to collect us at our present destination, we are willing to give up location data from our connected device. In the same way, some of us are willing to pay for things using a device – whether it’s a phone, a ring, or a car. Each of these actions requires a different level of trust.
When it comes to data use – trust will always be at the center – how much trust we allow, what we give up, where our information goes, who sees it, what’s being collected. As the application of data monetization emerges and evolves, no doubt there will be trade-offs.
To learn more, please watch the full panel discussion on Data Standards & Monetization here. If you’re interested in building your cognitive business with Watson, take a look at our website to kickstart your IoT strategy, or speak to a representative to find out how we can help you take the next step.