Roundtable: Using data to defend your territory

Recent months have seen online retail giant Amazon cast its gaze over the insurance sector, with job adverts touting plans to “significantly grow” its insurance business and even rumours of the group attempting to poach Lemonade’s staff.

Growing speculation that other tech giants such as Facebook, Apple, Microsoft and Google could make a play for the UK insurance customer will only be adding to the unease felt by some incumbents.

Against this backdrop, Post, in conjunction with Cognizant, invited industry leaders to discuss how insurers can make use of customer data to defend and build on their existing market position.

Daren Rudd, chief architect at Cognizant, began by asking the group whether insurers are making the best use of the data they already hold on their customers.

“Insurers talk about the history that they have, 100 years here and 100 years there, and obviously tonnes and tonnes of experience of working with customers,” he said.

“So I’m quite interested to understand why it is that when we work with insurers, we sometimes find they’re not making the best use of their data. They’ve got lots of silos, it’s stored everywhere, but they’re not really leveraging it.

“So if the history and the back data that you have on your customers is a competitive advantage, why aren’t you using it more?

“Is all the money and the effort that we put into building data warehouses, pulling all that data together and getting a single view of the client really worth all the effort if, actually, someone else can come in and look at a client and understand them better without actually knowing that client in the way that you do?”

Saga’s director of actuarial services Tony Richter replied: “It might be that we don’t make the best of it, you can always do better. And it’s massively valuable, both from an underwriting, risk pricing point of view, and also from a marketing point of view.”

The right data

Steve Jackson, head of financial crime at Covéa, said he wondered whether the right type of data was being collected by insurers to compete with companies like Amazon and Google.

“Traditionally, if you collect enough information, you are able to deliver a client service,” he said. “So we’re interested in whether or not this fits our criteria for a client service.

“Actually, this paints a picture. When you think of the likes of Amazon, Google and all these companies, they’re actually building a picture, an individual picture, about a person and their relationship with the surrounding friends and family and that kind of thing, to be able to consider what products might appeal.”

Steven Wilkins, head of underwriting insight at Hiscox, noted: “There’s a couple of different levels to the type of data you can collect and what you can use it for. I’ve always thought that the types of data which insurance companies have been collecting the last 10 years is actually quite useful in the way that you price.

“If you’re looking at small commercial insurance, you might be collecting turnover and wage roll and what they’re doing, and, actually, that’s a fairly good predictor of claims and what people might do.

“But when we come to interacting with the customers, knowing what we might find from their Facebook profile or who their friends are or where they are in their life, actually that’s better when we’re marketing and deciding how to offer products to people.

“And for me, that’s where incumbents and our traditional big insurers have fallen behind, in building products and strategies to work with customers, knowing what you can find from social media and elsewhere.”

Kristian Feldborg, co-founder of Policy Castle, asked whether insurers were acting quickly enough to discern which data was relevant to their existing customer bases. “Sometimes you ask the question: ‘How many circus artists do you actually have on your books?’ A lot of this data that you’ve gathered over 100 years is totally irrelevant by now, but it hasn’t been tidied up. It just stays there. And no one has looked forward. What is a normal family actually going to look like in 10 years’ time?

“We know what it looked like in the 1970s, because that’s exactly what we have on our questionnaire. But getting an insurer to stop looking backwards and then take a look at what the future’s going to look like is quite challenging.”

Axa’s director of commercial intermediary Deepak Soni said: “From my perspective, historical data that insurers hold is of real value, but it’s only of value if we can release that value. There’s different layers of that as well. Insurers have different journeys of enriching or looking at that data they already hold, they need to plug it into other sources of data.

“Whether that’s newfound data, using Facebook, using Linked In, using other streams that are available, that’s a next step. And I don’t know how many insurers are there, at that point. Some may well be there already; others might be trying to get to that point.”

US on-demand insurer Slice recently confirmed its plans to expand to the UK after expanding to 13 states across the US. Its director of operations Dawid Glawdzin said: “Our approach to data is different, because we don’t have any historical data. We operate on demand in the economy, so we had to create the data, but we had to do it in a new, innovative way as well.

“So, for example, in the US, with our short-term homeowner insurance for the Airbnb host, we ask two questions: what’s your name and your address? Because we can tap into publicly available databases and some third-party databases and we have everything we need to underwrite. So that’s a new, innovative approach.

“And we can provide cover just for the duration of the stay, and then it goes back to renewal. Where you have to wait 12 months to get a new datapoint, we’ve got it every couple of days. And we can put the data back into the process and make the product better.”

Measuring intent

Sean Heshmat, assistant vice-president of Cognizant, said: “Part of the challenge, when I look at outside industries, is that in insurance we really are not very good at measuring intent or propensity to do something.

“If I, for example, just bought a house and I go from a rental property to a purchased property, there is an intent for me to cover my house with probably a lot more content insurance.

“That information is available out there, because I’ve probably tweeted or probably said something about buying a new house. But the chain of getting that information at an insurance company to action it is really slow.

“Basically, if I went into a home insurance comparison website and I’ve actually then analysed the market, maybe in some of your systems you have that information that I maybe have browsed different home insurance calculators, for example.

“The real issue is we’re still spending a lot of money because we don’t understand the intent, the propensity and the timing, because once I’ve made the purchase, I do not need to have all these banner ads plastered all over my websites

“The worst thing is I’ve actually bought it from you, and two or three weeks later you’re still plastering me with ads saying ‘buy home insurance from me’. That really demonstrates that we’re not joining this data up.”

Chris Wyard, head of underwriting data at Allianz Insurance, agreed: “The customer expectation has changed, and we can all recognise that, but the reality is: are insurers, or is the industry more generally, really harmonising and harnessing the data that they have across all of their business?

“Have we developed a view on customer that is applied consistently across pricing, across distribution, across our claims proposition, across our product development? Because if you don’t, you have that disjointed nature.”

David Pearce, head of pricing at Aviva, said: “There’s a perception that we’re good at risk analytics. We aren’t. There’s actually a hell of a lot more we can go towards. And as an insurance industry, new technology through machine-learning will actually give us massive uplifts in our risk modelling.”

Impact of telematics

Meghan Anzelc, chief analytics officer at Axis, considered the impact of telematics and the data it has been able to provide to motor insurers. “The example that I consider is telematics, and particularly with all of the aggressive investment in that arena in the US. They sort of presumed that it was going to be valuable, and put huge investment in it, and didn’t see a return on that investment venture for years.

“But they’ve got trillions of miles of information, to the point where, when they changed the product a handful of years ago, they said: ‘We don’t even want you to keep the device more than three months’, which now means their device costs are managed. Acquisition of additional data is not their primary goal. They now know what the value is.”

Rudd asked: “Do you think there’s a competitive threat to you, as insurers, that when we look at where those devices are and who owns the actual data source, it isn’t the insurer?”

Andrew Harley, global leader of the advanced analytics practice at Willis Towers Watson, believed so. “Yes, I’d say that is a massive risk for insurance companies,” he said. “The data that we were talking about earlier, propensity data, that’s a threat in the context of distribution.

“So Amazon and Google are extremely well-placed, should they wish to, to understand propensity behaviour and place better ads and essentially drive new business. But they don’t, at least to this point, understand the risk.

“The data you’re talking about there is fundamentally risk data, and insurers should be worried about companies that manufacture devices, and car manufacturers that fit those telematics devices in the cars. Because the power is going to be then, it’s as simple as that, they will know the risk after three months.

“So we need to distinguish between the types of data that we’re talking about when we think about competitive threat. I agree that the data that insurers have already in relation to risk is a competitive advantage.

“It’s not a barrier to entry. Frankly, it depends which product we’re talking about. From a commercial insurance perspective, probably the experience of the underwriter is just as important as historical data.”