McKinsey: Insurers are aware that they are in the midst of a huge disruption. What do they need to do to take advantage of it?
Stefan Heck: I think they need to make three big changes, and these need to take place concurrently and quickly. First, those that thrive will rethink the product. They will move from a model based on a once-a-year transaction with the customer, an interaction when a loss occurs, and minimization of the payout, to a model where the company actively engages the customer the way a Facebook or a Google would.
The second transition concerns the talent base and the kind of skills and capabilities needed. Historically, insurers have focused on backward-looking data, on actuarial science, on estimating future risk from past performance. If motor vehicles undergo massive change so that they don’t drive the same way and accidents are automatically prevented, then historical data will not be predictive of what’s going on. We see this already with government data: the latest available US government statistics do not feature Uber and Lift, so cannot be predictive of what’s happening right now in some leading cities where these new models have high usage rates. Therefore the actuarial science will be replaced by data science using forward-looking, predictive analytics based on real-time data. That requires different types of skills.
The third transition will take place in business processes. Today, we have a relatively linear, and for many insurance companies relatively uniform, process for buying insurance. But in the digital world, I will pick up my cell phone, snap four pictures of my car, snap a picture of the license plate, of my driver’s license, of my odometer, and boom, I will get an insurance quote, I will get competitive bids. They might be priced on mileage, or on whether I drive in the rush hour, or on whether I drive for Uber. But it will be a fully digital, mobile-centric process.
Similarly for claims. Today I call up the insurer, I have to fill out an amount of paper work, there might be a visit to assess the loss, and I go to a repair shop. Those are now more digitally hooked up so they’re feeding data into the insurer, but it’s still after the fact. In future, the insurance company will have to be able to process claims data in real time. At Nauto we are uploading instant video of what happens inside and outside a crashed car. We can see who got injured, who’s walking away, what happened to the car. That information won’t come from a questionnaire anymore, it will be live feed data coming in within seconds of an accident.
And if the claim is a simple fender bender—rather than a more serious injury claim that requires hand-holding on the part of the carrier—then that claim process will become fully automated. I have not yet met an insurance company that’s comfortable fully automating claims end to end so that a machine-learning algorithm will decide how to settle a claim, but that’s where it’s headed, certainly for simpler claims.
It’s a model that has a huge advantage in terms of cost structure. If you picture a traditional paper-based, IT-heavy, serial process competing with this very closed, all-digital, all-mobile loop—that won’t be a fair playing field to compete on.
That all amounts to gut-wrenching change: changing the offering, the talent, and the core processes. We have seen companies win on a huge scale in these kinds of transitions. If you go back 100 years to early industrialization, people who manufactured in an assembly line, who were able to standardize the product and to create global marketing channels and operations, are the leading companies that we know today. The same kind of transition is coming now for the innovators who can take advantage of this once in three- or four-generation change. But some brand names will inevitably disappear.
The case for digital reinvention
McKinsey: Tell us more about car insurance. What has been the impact of digital on the business model to date, and how is it evolving?
Stefan Heck: We are moving to usage-based insurance, but it is still very much in its infancy. The current model offers one-step improvement, using real driving behavior for underwriting and pricing decisions instead of relatively generic details such as zip code, resident address, and years of driving experience, so you get a more precise picture of who is a good driver and who is a bad driver. But the pitch so far has been about discounting premiums rather than improved safety and services. So that’s why I say it’s still in its infancy. There’s a lot more opportunity to actually help drivers and make them better, not just to rate them and score them.
For example, I can get more information as cars start to build in video, traction-control sensors, and other devices. At Nauto, we retro-fit machine vision and cameras so that you can actually see what’s going on as opposed to having to infer what happened just from braking and acceleration behavior. What’s the traffic flow? What are the weather conditions?
Usage-based insurance so far has been based on two relatively simple models, either pricing just for the mileage, or pricing based on how often you brake or make severe vehicle movements. But actually what you want to price for is the full richness of when you are driving. Who is driving? How are they driving? Are there any other passengers or other risks in the vehicle? And then how many of the events that happen are due to external environment? Driving in a busy city during rush hour is different from driving erratically on a country road where there’s no other traffic.
What we’re looking at in future is a usage-based offering that not only rates you more accurately but also reduces the risk while you’re driving. We’re getting into warning features, into the ability to intervene and provide help when incidents happen.
So we’re headed toward a world where, similar to a sports game, you can hit instant replay and say, “Let me study what happened in this accident. Where were the vehicles? Which way were they going? What were the speeds? Who was distracted or not? What was the state of the light? Who ran the stop sign? And how did the accident unfold?” I can then attribute fault in just a few seconds of instant replay, but I can also say, “Did the airbags deploy? Do I need to call an ambulance? Do we need to call a tow truck?”
All of that data is now available in real time from sensors and can be embedded into my claims process to reduce the costs of repairing the vehicle or of medical treatment that might result. And that begins to open the door to a very different relationship with the customer, who can feel protected and cared for by their insurance company as opposed to the experience of a once-a-year interaction where they renew their policy and may or may not get a premium adjustment.
We see some carriers that understand this is the beginning of a reinvention of the insurance model from a risk pooling to a prevention and service business, but we also see many carriers that are still scared of the technology, a bit like the utility world was a few years ago where people said, “You know what, I’m fine running my coal plants, I don’t want to know about all this renewable technology because it’s only going to hit in the next generation after I retire.”
But car makers are adopting the technology quite rapidly, particularly at the luxury end. In five years they won’t be retrofitting devices any more. Devices will be built in. And five to ten years out we’re going to see some very, very major effects.
We’ll see a dramatic reduction in accidents as real-time collision warning and increasing automation come into vehicles—by 70 or 80 percent in the long term. At the same time, the severity of some of those accidents will go up because there’ll be more electronics and more value in the car; therefore they will be more expensive to repair in terms of replacing electronics modules, for example, so that will offset some of the frequency dimension. But it’s a very different sized market if we think about 70 percent of loss events disappearing over the span of ten years or so.
And we’re going to see real effects because the fixed costs for most insurance carriers are still quite a large portion of the cost structure. If the variable loss amount drops many companies will have to reinvent the business model or face dramatic profitability consequences. As this transition occurs, we’re going to see more of a spread between the innovators—those that take advantage of the new models, create new insurance products, and get more deeply connected to their customers so they can have an ongoing relationship in terms of services, protection, and convenience—and those that are late to these transitions and that drop in value.