Pricing is key to this. Companies will not be able to connect with customers, grow market share and deliver sustainable margins unless they have systems that can land a price that connects with customers and edges out the competition. And, said Michael O’Connell, Chief Analytics Officer at TIBCO, this increasingly means investing in a dynamic, real-time machine learning platform that is responsive to interactions with the software.

“The secret sauce is do people accept or reject a quote and as this data accumulates, you rebase your model to respond to how customers are behaving in different product areas”

 

Michael O’Connell, TIBCO

“Underlying that you have to have modelling for underwriting risks, customer conversion and retention, and price optimisation for profitability,” said O’Connell, speaking from San Francisco. “You need a number of machine learning models in the platform, including price discount tables based on microsegments such as customer tenure, business goals, customer history. And the secret sauce is do people accept or reject quote and as this data accumulates, you rebase that model to respond to how customers are behaving in different product areas. And you need to do all of this really quickly.”

There’s no doubt that machine learning and new analytics tools promise to revolutionise pricing strategies and product innovation. But Antton Peña of Flock Cover said delivering customer-focused innovation isn’t just about the latest technology. “There’s tons of technology that’s been around for 10-20 years that still has a lot to offer,” he said. “We’re still not making best use of Internet of Things and cloud computing to solve the problems around us.”

Michael O’Connell of TIBCO said he found the companies that had been successful were those that focused on customer lifetime value. “How do you convert and retain the customer based on historic data and quote acceptance?” he asked. “What’s the pricing elasticity of different customer segments to discounting? Apply the data with the lens of lifetime value to the interactions you have on a daily basis.”

“How do you put together a backend that has that 360-degree view to get the best data into the predictive models in this dynamically changing data landscape?”

 

Michael O’Connell, TIBCO

He acknowledged that the underlying data science is very sophisticated and there were systems challenges in managing data from a multitude of different sources. “How do you put together a backend that has that 360-degree view to get the best data into the predictive models in this dynamically changing data landscape?”

“Even small insurers can get into the market if the data sources are broad enough.”

 

Jon Holtan, Fremtid

Jon Holtan of Fremtind pointed out that the growth in the data universe created openings for start-ups, eroding the data headstart of incumbents. “Many years ago it was the number of data observations that was considered important but now it’s more about the number of data sources,” he said. “The world of your data is much more important now and this means even small insurers can get into the market if the data sources are broad enough.”

Dynamic pricing: pros and cons

“The link with profitability is not ultra clear to me.”

 

Frederik Borgers, UNIQA

The jury is still out on the value of dynamic pricing.  While in theory it should allow a company to undercut competitors and squeeze market share, Frederik Borgers of UNIQA said “the link with profitability is not ultra clear to me.”

“For us, dynamic pricing…is about being more fair and creating flexible products for our customers.”

 

Antton Peña, Flock Cover

However, Antton Peña of Flock Cover said he was a big fan of dynamic pricing. “For us, dynamic pricing is not just about competing with our competitors,” he said. “It’s about being more fair and creating flexible products for our customers.”

He said much of this advantage comes down to having the right data and being able to accurately assess risk at a very segmented level. “When the customer becomes safer, they end up generating profit for the insurance company, which then bounces back to the customer through better pricing, so it’s a win win,” he said.

Michael O’Connell of TIBCO agreed. “Dynamic pricing isn’t about duking it out with everyone online for the lowest price. It’s about the life time value of the customer and putting a stake in the ground about what you want to get for the premium, what your underwriting risks is, what the customer lifetime value is and what’s the sensitivity to discounts. There’s a sweet spot in all that for the products you have and the premium you want to target for profitability.”

He said this was a continuous process, of endlessly rebasing the models to see how they are playing in different markets.

These systems are not cheap, of course. However, Michael O’Connell of TIBCO pointed out that many companies are already shouldering the costs of old black box systems that are, he said, “ripe for the taking”.

“People do not have control over them, they’re inflexible and expensive so you can justify the cost of implementing more dynamic real-time analytics platform because you can then take out the costs of the old system,” he said. Then it’s a case of measuring the impact of the platform’s pricing outputs on product uptake and profitability in different markets. “You have to measure it in a fairly detailed manner,” he said.

Profitability isn’t just about top line growth through accelerated sales when products hit a pricing sweet spot. O’Connell said these kinds of analytics can also identify cross-sell opportunities and help identify fraud. “You get a more complete ROI picture when you pull all of this together,” he said.

“Fairness, transparency and flexibility are the three key pillars for me”

 

Antton Peña, Flock Cover

Data ethics

As always with AI and machine learning, it was also important to consider the ethics of the technology. “Fairness, transparency and flexibility are the three key pillars for me,” said Antton Peña of Flock Cover. He stressed that transparency helps customers understand the value they get as a result of sharing their data with the insurer and that they benefit from the insights generated from the system as it helps them mitigate their risk exposure.

“Explain-ability is really important”

 

Michael O’Connell, TIBCO

Michael O’Connell of TIBCO agreed. “Explain-ability is really important,” he said. Jon Holtan of Fremtind agreed, saying the data ethics around AI is going to become more and more important for insurance companies.

Antton Peña of Flock Cover said the nub of the issue is to give back to the customer. “Do not just get their data to better your own pricing, give something back to the customer.”

“What’s good for the customer is good for the insurer.”

 

Jon Holtan, Fremtind

Jon Holtan of Fremtind agreed. “What’s good for the customer is good for the insurer.”