Huge volumes of data make real-time insurance a possibility

Article tiré du magazine The Economist.

On-demand insurance becomes possible

Even at weddings or whale watches, the buzz of a drone is no longer a surprise. Drone photography is booming. Gartner, a consultancy, says some 174,000 drones will be sold for commercial use around the world this year, and 2.8m to consumers. It is easy to imagine a few might fall out of the sky, causing damage the pilot cannot hope to pay for: crushed wedding cakes, injured spectators and so on. Amid scores of near-misses, several incidents have already occurred. In 2014, for example, a drone filming a triathlon in Australia crashed on a competitor’s head.

Clearly, drone-users need insurance. Typically, risks are insured through the payment of an annual premium. Insure4drones, a British specialist, charges £738.86 ($1,000) to cover a DJI Phantom, a bestselling drone, for a year. From October Flock, a London startup, will offer insurance on a flight-by-flight basis, at the push of a button in an app, to any commercial drone-operator in Britain. Cover for amateur pilots will soon follow. Costs will be about £5 per hour of flight, according to Allianz, an underwriter.

Flock’s app relies on a wide range of data. Weather forecasts come from IBM, a computing giant which, having spent over $2bn on The Weather Company in 2015, now offers forecasts to within a few hundred metres, and over a period of minutes. Live information about nearby aircraft is provided by a software company called Snowflake, which tracks aeroplanes around the planet. Flock also considers local topography, such as proximity to churches, hospitals and schools, as well as roads and traffic levels. It also monitors the drone itself, gathering data as it flies to build a risk profile for that machine. All these numbers are crunched when a customer requests insurance through the app. As well as offering a quote, the app tells pilots how to reduce their risks.

Allianz then converts Flock’s data-driven risk scores into a price. The attraction for Allianz is acquiring customers cheaply. “Rather than humans sitting and writing business, the algorithm does it on the spot,” says Tom Chamberlain, who manages its aviation underwriting.

Conventional insurance works by pooling individual risks and then setting a price for that group—new drivers under 30, say. But that process can be much refined if the objects and people being insured can report to the insurer automatically, and if there is a wealth of data on the external environment. As an ever-growing number of sensors—in phones or watches, drones or cars—gather ever-greater volumes of data, more and more activities can be assessed for real-time risk (though in the absence of pooling, some risks may become prohibitively expensive to insure).

Flock is not alone. Verifly, a New York startup, competes with it in America. Root, a car insurer, offers drivers insurance based on their minute-to-minute behaviour behind the wheel. It even offers a discount to Tesla drivers if their car spends plenty of time in autonomous mode. Slice, a San Francisco startup, lets its customers insure their houses and cars for the time they are used on services such as Uber and Airbnb. Trov, also from San Francisco, insures personal possessions for short periods.

Flock’s chief executive, Ed Klinger, says that he eventually wants to insure all kinds of future autonomous activities, from taxi rides to rolling delivery pods. He argues that selling insurance through annual premiums is inflexible. It less easily takes advantage of the large volume of live data that can now help estimate the risk posed by a given activity at a given time. For instance, a passenger in an autonomous taxi may be at far lower risk if the trip takes place outside rush hour, or in weather conditions in which the car performs at its best. Firms that dispatch delivery drones might use Flock to calculate the risk for each flight automatically, depending on cargo and address.

The business model is in its infancy, but on-demand insurance seems bound to grow. In a world where consumers expect push-button convenience from their services, they will demand the same of the insurance those services rely on.

 

From the print edition