man-using-digital-tablet

Big data and AI in insurance, for assurance

Estimated reading time: 4 minutes

Andrew Yeoman, CEO of Concirrus, opined that nothing can happen without insurance. “It is an essential service. We couldn’t have this interview call without insurance because it requires satellites to be put up in space, and cables be put under the sea, and the computers we are both on… none of these things could have been manufactured or put together, if it wasn’t for insurance.”

As a technology company that that automates underwriting, Concirrus recognises the challenge the insurance industry faces at being efficient when it comes to writing risk.

“Now, we are living in an age where actually, from an insurance perspective, it’s not that you can’t know the information, but rather you don’t have the right tools to know the information.”

“The science that sits behind the insurance industry, is scarcity of data – when insuring vehicles, you never know which would be in an accident and so on.”

Andrew explained all this changed about five years ago when thanks to the Internet of Things (IoT) and sensors being attached to assets, there was massive growth in data being generated.

“Now, we are living in an age where actually, from an insurance perspective, it’s not that you can’t know the information, but rather you don’t have the right tools to know the information.”

Pricing and selecting risk more accurately

When machine learning (ML) and artificial intelligence (AI) technology is applied to all of that data generated by assets, Concirrus found that they have been able to see how, where, and when assets are being used.

Behaviour of an asset’s movement, turned out to be a better rating factor or better factor at predicting claims than traditional static factors like the age of an asset, or where it was built.

If five years ago, there had been a lot of rejection of technologies to underwrite risk, today Andrew said they have been able to prove that their algorithm models outperform traditional models.

“When assessing driving risk, if you were to find ten people my age, my background, in my demographic…using traditional models we look like the same risk on paper.  But what if I actually drive to the train station and leave my car there, while the others were not?

Behaviour of an asset’s movement, turned out to be a better rating factor or better factor at predicting claims than traditional static factors like the age of an asset, or where it was built.

“So, that principle holds true in every line of business. If I can see how the risk is behaving, I can start to rate it differently.”

Whole product lines for the marine and transport industries are  built upon this Quest platform which accesses wide-ranging datasets and integrates them with historical information. This is to reveal asset behaviours that correlate with claims and enable more accurate quantification of risks to these assets that have been insured.

Ensuring that insurers put their money behind the right risks at the right price, is capital effectiveness which Concirrus enables with their algorithmic assessment engine.

When capital effectiveness is combined with operating efficiency; operating efficiency being what Andrew described as doing all of the above in the most automated fashion possible; the insurer can save time and money. This is possible via the Quest submission system, whereby Quest is Concirrus’ big data and machine learning platform.

Whole product lines for the marine and transport industries are  built upon this Quest platform which accesses wide-ranging datasets and integrates them with historical information. This is to reveal asset behaviours that correlate with claims and enable more accurate quantification of risks to these assets that have been insured.

With the same amount of rigour and diligence, an underwriter could potentially assess up to 50 risks a day, instead of just five.

Predictive maintenance

“We change the underwriter’s time from being a retrospective assessment; or looking at things in the past, to being a prospective assessment, or what might happen in the future,” Andrew said.

Instead of trying to simulate a future from past behaviour to quantify risk, underwriters can now make informed decisions to mitigate that possible future risk.

“This whole notion of assurance is a huge change that is coming through in the industry, “Andrew opined.

That alert is something that insurers have never ever had the data to do before. “And they don’t have the organisation or ecosystem to support that. But now they have the data and they need to develop the response.

Imagine if the plaster ceiling of your house was going to fall in because of moisture build up.  If an IoT sensor had detected that, the house owner could be alerted and the plaster ceiling falling could be prevented.

That alert is something that insurers have never ever had the data to do before. “And they don’t have the organisation or ecosystem to support that. But now they have the data and they need to develop the response.

“We are heading towards that future where what was unknown, is now known. What was unknowable, is now unknowable .”

So instead of insurance, assurance is actually now possible. And Andrew observes that it is absolutely happening, only not at scale.

He concluded that insurers need to learn how to operate in this automated world with all of its massive amounts of data.