For insurers and technology partners, this is a fun question to ponder: How smart can insurance get? Perhaps an even broader question might be: “What is smart insurance?” What does it look like to apply analytics-based decisions to the process — from underwriting through claims? More importantly, what does it look like to apply penetrating data knowledge to individual people and individual risks?
I think these answers may lie in a closer look at our human relationships, and how they closely parallel what insurers are trying to do with a wide and growing array of risks. As the insurance industry shifts its concerns, adds digital connectivity and mature data analytics to its portfolio, it may come to look, sound and act much more like your mother.
After you’re done thinking about that picture, let’s consider it a moment. Insurance technology is striving to become cognitive and connected. The cognitive
part will be forecasting problematic issues and preventing claims events. It will be asking who your friends are and wondering where you hang out. It will seem like it cares about you, and in some ways it will. The connected
part will be deriving relevant insights from everywhere.
See also: 4 Steps to Ease Data Migration
“Smart insurance” will be insurance that knows its insureds well, and the insurers that survive and thrive will be INCREDIBLY smart, powerful and successful.
The only way insurance will become smart, however, is through data. Data is the gatekeeper to future insurer success.
The long-term competitive advantages to be found in data will be found by those who are collecting data across long periods. Data is like calculus or learning a foreign language. It is a building-block science that requires hands-on learning and manipulation to grow its usefulness.
Insurers that are dealing with data well today, are going to have a long-term advantage.
To illustrate my point, I’d like to look at three aspects of data that we will probably be thinking about for however long insurance is in existence. These three are: patterns, volume and experience. All three play into an insurer’s data capabilities.
Analytics is all about patterns or lack of patterns — finding the signal in the noise.
In one of my previous blogs, I discussed my affinity for Pandora. Just as my Mom could tell you my first word, Pandora can tell me the first song that I ever listened to on its service. With every song I listen to, it learns more about me. We’ve grown close. It knows what I like and what I don’t, so it is able to identify the signal data and tune out the noise data. How does it do this? It takes my personal data and cross-references it with its 100 million other users to find patterns.
As amazing as that is, pattern analysis in insurance has far greater implications and far more exciting applications. With it, we’ll be able to home in on signal indicators within the data and tune out the noise, identifying what’s unnecessary. This will result in an insurer’s ability to make “on the fly” decisions based on patterns that have been learned through cognitive systems, such as IBM’s Watson. Recently, IBM and Majesco announced a partnership (you can read the press release here
) to bring cognitive capabilities to cloud insurance offerings and insurance capabilities into the cognitive sphere. Data gives a cognitive learning system the food it needs to accomplish well-rounded learning and growth. The more relevant data it can consume, the better it can find patterns and separate good risks from bad.
Data volume is a crucial aspect of the long-term data advantage.
While some companies worry that they have too much data to structure, organize and store effectively, many simply don’t have enough. They are either letting data streams sift through their fingers like sand, or they are not seeking the relevant data streams that will empower their risk selections.
When they are thinking of data, they may be thinking about the three or four traditional data sources that normally point to good risks or bad. In underwriting, for example, a common point for data scoring, insurers may only pull from a few common sources for information on applicant history.
Yet, the future of data decisions may look more like Mom than we know — weighing the big picture and all of the little details. There may come a point where insurance companies shy away from questionable risks on a sort of “data-formulated hunch,” based not on any one large factor but on a hundred tiny hints. Applicants with previous similar profiles turned out to be bad risks for no apparent reason. Maybe we’ll call it insurance intuition
. But insurance intuition will only be possible with large volumes of long-term histories, combined with relevant real-time data streams. The difference between insurance decisions yesterday and those of the future will look like the difference between a handwritten description and a hologram. Insurers are beginning to crave the transparency that data can provide.
To prepare, insurers need a well-planned and well-structured data organization. They need definitive data knowledge across the enterprise, knowing where they are generating data and which data streams are currently being used. How is the data structured for usability? How is the organization archiving the data for later use? Then insurers need an understanding of what new data streams may exist outside the organization that will add value to their analytics. All of these considerations require insurers to continuously build their volume of usable data.
Experience unlocks data’s long-term value.
Insurance is about experiences. The more experiences that insurers can record and analyze, the better they will be positioned to accept risks. But the future of experiences and modeling likely outcomes is so much more than that. For an excellent example, let’s look at Google’s work with autonomous vehicles.
Google can’t just place a car on the road and let it drive. It needs the system to learn about hazards, driver behavior, traffic patterns and sensing the unexpected. It needs millions of hours of experiential data — far more data than it can acquire with daily driving. What Google has done, is to use real data as the seed for simulations. These simulations model thousands of possible outcomes to any given situation, “teaching” and rewriting the software to adapt without road time. In this way, the Google car is gaining experiences without experience.
See also: 3 Types of Data for Personalization
Think of what insurers could do with similar simulations. Using experiences to build new experiences and model thousands of different outcomes to the same event will make insurers better equipped to predict, prevent and protect their policyholders over the long term. As insureds approach a likely claims scenario, data’s cognitive déjà vu will kick in and avert a claims event. For insurance to grow smarter, it needs to reframe what it means to model scenarios based on experience.
Experience of a different kind is also a key factor in data’s long-term value. Insurers simply need time to grow their data mastery. Analytics requires testing and validation. Experience, as well as tools, approach and data sources, is what will allow insurers to mine the best analytics from the data they own.
There is no time like now.
Now is related to the future. It’s the future’s history. If you would like to build an effective data organization or plan your company’s vital data strategy, there is no time like now.