Four steps insurers must take to realize the potential of IoT
The number of IoT devices on the planet has now surpassed the number of smartphones. After half a decade of hype, insurers had expected to be writing "connected policies" by now; rather than continuing to rely on antiquated data and one-size-fits-all plans, insurers were looking forward to creating targeted insurance policies with real-time data, unique to each customer’s needs and risks. The promise of IoT is wide-ranging: more accurately priced business, home, and auto insurance, as well as personalized advice to prevent damaging events, could save customers and insurers from catastrophic losses.
However, most of this has yet to happen... so what gives?
The Phases of IoT Adoption
First, it’s important to note the first wave of IoT and data adoption is not yet complete. In fact, there are five distinct phases over which the full value of IoT will be realized.
The first few phases are really about getting the end customer or consumer to do their part in adopting IoT devices. Phase one is simply getting the device into the hands of the customer. In the insurance use case, the technology company may use an insurer as the channel through which to sell to the end customer and as a provider of subsidies. For example, Canary partnered with State Farm, while Nest partnered with Liberty Mutual. Phase two, meanwhile, consists of the customer installing the device and agreeing to share the data with the tech company and the insurer.
Phases one and two have been happening over the last several years.
In phase three, the customer receives real-time notifications about conditions at their property and may choose to respond to those notifications with behavioral changes. Phase three is happening today. Companies like Notion and Pillar are selling and installing their devices in homes and construction sites, respectively, across the United States, and providing information that the customer can react to.
Next, in phases four and five, the insurer needs to wrangle the data generated by these IoT devices and then make sense of it. Phase four is the collection of device and user-generated data by the insurer in a structured and organized way. Phase five of insurer IoT adoption is the process of applying AI and machine learning technologies to this data in order to derive new insights that enable better underwriting, more accurate claim resolution, and the creation of real-time, connected policies.
Phases four and five are only just beginning to take shape.
Until phase five, insurers are like The Sharper Image, selling and testing a slew of novel gadgets for guinea pig customers. But that does not mean these phases are without their uses. In the early days of IoT, Insurers have been piggybacking on customers’ willingness to adopt technology and respond to notifications. This willingness has led to positive risk selection for insurers, positive changes in customer behavior, and a sharing of responsibility on the part of the customer for decreasing risk (rather than just expecting to be indemnified for losses).
But what needs to be done to get to phases four and five so insurers can fully harness the potential of IoT?
Startups must be better at delivering data that is usable by insurers, not just raw numbers."
There are four specific obstacles that need to be addressed:
The BIG secret in insurance is that insurers are actually terrible at using their existing data. The real value of customer data comes in combining it with existing insurance company data, but insurers' internal data is ALL over the place. It needs to be extracted across different systems, checked for accuracy and consistency, and then integrated with external data.
Insurers need to get on the same page internally. Departments are siloed and marketing, underwriting, and claims do not talk to each other. IoT startups establish a relationship with one department and get stuck there. The marketing department doesn’t reach out to underwriting and say “hey we are having success with this IoT device; are you interested in using it to improve underwriting?” The problem is that each department has different goals: for marketing, success is an increase in lead conversion when IoT device is offered at the time of sale. For underwriting, success is achieving positive selection and loss avoidance. For claims, success is paying fewer fraudulent claims, and paying the rest faster.
Increasing Data Capacity
Lastly, Insurers don’t have a process to ingest data from the 34B IoT devices that will be in place by 2020. Here, insurers need more help from IoT startups. Startups must be better at delivering data that is usable by insurers, not just raw numbers. This entails having a deep understanding of actuarial algorithms and the inputs that are used to create and price policies so the right data can be ported into existing workflows.
Facebook, Google, and Amazon are starving other industries of AI talent. In order to glean insights from data, insurers need machine learning experts, but the technology is changing quickly and the talent pool is concentrated, with many experts working outside the insurance industry. And, regardless, AI is not quite there yet. Below is an example of Google’s image recognition engine mistaking a turtle for a rifle. It also mistook a baseball for espresso.
There's still a lot to be done, but the tipping point for IoT is not as far as it seems. By 2020, insurers should be well positioned to overcome these obstacles and derive the insights they need to precipitate large-scale change in how they do business.
The early days of IoT were all about customers’ willingness to adopt new technologies and change their behavior. The next stage of IoT in insurance will be about insurers changing THEIR behavior. This includes changes in underwriting practices, in claims processes, in new product development and in methods for storing and using new kinds of data. Once these changes are in place can connected insurance policies become a reality. Only then will IoT devices become a cornerstone of insurer policy creation and administration.