Hardly a day goes by without a news story about data science or artificial intelligence in the insurance press – but what does a data scientist actually do? How are insurers using data science to improve customer experiences, propositions and pricing? And how can you get the most from data science in your organisation?
Greg Brown talked to Orlando Machado, Aviva’s Director of Customer Analytics and Data Science, to get some answers.
Greg: Orlando, how did you get into data science in insurance?
Orlando: After completing a PhD in statistics at Warwick university I started work as a statistician in academia. And I’ve been working with data ever since – I’ve run data and analytics teams for various data-centric businesses such as Wunderman, part of WPP; dunnhumby, the analytics business at the heart of Tesco Clubcard, and was most recently Chief Data Scientist at MoneySupermarket. Aviva is my first stint with an insurance company.
Greg: What were your first impressions of insurance compared to other industries?
Orlando: When I arrived in the insurance industry, I observed that insurers use data extensively but only in a few areas like pricing and risk. They are far behind the curve when it comes to using data for, say, propositions and marketing. When I’ve been recruiting my team, I’ve therefore been looking outside the industry.
Greg: What are you focusing on now?
Orlando: Now that we have the makings of a team our next job is to build a customer segmentation. Although segmentation has been carried out by businesses for many years, we use a number of contemporary machine learning techniques that give us a comprehensive way to classify our customers. This will allow us to design new propositions targeted at specific segments and deliver targeted marketing across broadcast media and social channels. To do this we’re tagging attributes against each and every customer. For example, we might tag individuals with a digital engagement score and use this to target our digital marketing spend. We’ll also use this data to help us make better product recommendations within MyAviva, our digital customer portal.
Greg: Where do you see data science unlocking the most value for insurers?
Orlando: That’s a very big question. Here’s a few areas where I see a big opportunity. Firstly, claims – Aviva is doing a lot of work using artificial intelligence to identify and eliminate invalid and fraudulent claims. I believe there is a big opportunity to automate claim approval and payment at the point of FNOL, which reduces costs for us and improves experience for the customer.
Secondly, connected and self-driving cars. The volume of data produced by connected vehicles is only going to increase, providing both an opportunity and a threat for insurers. If insurers can position themselves in the value-chain and get access to the data, there is an opportunity to create highly tailored insurance products and pricing. The risk to the industry is that manufacturers disintermediate insurers – Tesla recently announced that they would insure all of their cars forever. [Link to Tech Crunch]
Finally, and more in the life space, genomic data. What happens when you can use data to very accurately predict an individual’s likelihood of suffering from certain diseases…
Greg: Those opportunities all lead to highly personalised insurance propositions. Do you worry about insurers becoming so “smart” that there is no longer any risk pooling?
Orlando: This is certainly an important issue, and highlights how data science can create new ethical questions. The risk is that those who need insurance most are priced out. No-one yet knows how close we will get to that scenario. For motor and home I believe that there is enough randomness in the system – you’ll never be able to predict claim likelihood with sufficient certainty. For life products this is potentially a bigger risk as you can predict with much more certainty, and as an industry that we don’t currently require anyone to carry out genetic testing before buying certain products. We are always keen to have an open dialogue with regulators about these topics.
Greg: How do you see GDPR affecting what you do?
Orlando: If anything, this is an opportunity for Aviva. We have always been rigorous about how we protect and use customer data, which puts us in a very good position. I also welcome the principles of transparency for the customer. This will hopefully help grow trust. [See a primer on GDPR here.]
Greg: What are some of the challenges for insurers in building data science capabilities?
Orlando: Recruitment is a big challenge. It’s very difficult to recruit new skills like data scientists. Traditional insurers – and many other businesses – just don’t know what good looks like and often make mistakes due to this ignorance. I have seen candidates walk into well paid roles on the basis of a number of technical terms on their CV but with little or no practical data science experience. Insurers need to admit to this lack of knowledge and recruit carefully to avoid building teams that talk the talk but don’t walk the walk.
Also, there’s communication. It’s really important that data science teams can collaborate well with other teams in the business – this means that communication skills and commercial awareness are crucial.