Traditionally, a competitive edge in the insurance industry has been created by differentiating on premium. Therefore, many strategies are focused on premium/price differentiation and this is reflected in the ways Data Science has been implemented in the industry nowadays. But aside from premium differentiation, there are other ways to differentiate as an insurer.
The insurance market is a price-driven market, because of the transparency of premiums in online markets. A consumer can visit for example Independer.nl or Comparethemarket.com to get an overview of different premiums from different suppliers for his/her car, health or life insurance very easily. The consumer fills out his demographics and coverage preferences and selects the best alternative, often focusing on top 3 positions only. The consumer’s choice highly depends on the premium that has to be paid. Therefore, Data Science in insurance is mostly applied to offer the best price for each individual. But this is not the only way insurers can benefit from Data Science and Machine Learning practices!
Example of transparency of premiums in online markets
In this article, we describe three other ways Data Science and Machine Learning are enabling insurance companies to differentiate from competitors in the insurance industry. The three practices evolve around the topics of product personalization, improved customer service and insurance claim prevention.
When a customer is able to adjust the qualities of their insurance policies to perfectly match his/her individual needs, we speak of product personalization. We will discuss two new ways how Data Science is enabling insurance companies to personalize their products. First, let’s cover level of excess, a promising feature to optimize since it interacts with the premium price in terms of risk. Using Data Science, we can predict the optimal level of excess and a better fit between a product and the needs of a customer. It gives customers more freedom to choose the level of excess that matches their personal situation. Currently the level of excess is determined by default options, but the level of excess you choose can have a significant effect on the premium you pay. Deciding which level of excess to take can be complicated to the average consumer. Data Science can help recommend the ideal level of excess for each individual customer, which makes a complex product as an insurance understandable to consumers.
Another way to achieve a better fit is coverage personalization. Based on what one’s peers might have needed in the past, we can recommend what a consumer might or might not need to have covered. what a consumer might needs to and might not needs to cover. This not only results in more cross-selling, but the customer also enjoys the feeling of being understood and have a secure feeling of being insured according his or her individual situation.
No one likes being ill or being involved in a car accident. Therefore, most communication touchpoints a customer has with his or her insurer are not the most pleasant ones. A health insurance provider can relieve the customer of some of this pain by providing an outstanding customer service. Offering some piece of mind in these dark situations could make the difference between a loyal customer and a churner. The good news is; with Data Science there are endless possibilities to create an enhanced customer experience, we will point out some evident ones.
The process of filing claims can be simplified by applying new ways to file a claim. For instance, by giving the customer the option to submit a photo or a video of the damage caused by an accident, after which an image recognition algorithm trained on damages, recognizes the situation and automatically estimates the costs of repair and subsequently files a claim. This faster and easier claim filing method will not only benefit the customer, but also the insurance provider, by improving accuracy and efficiency of their payouts.
Example of innovative claim filling
Optimizing the efficiency and effectivity of customer service centers also increase customer satisfaction. Using text analysis on written claims or notes from a claim filed over the phone, Data Science can discover the topic and connect you to the most relevant call-center employee or redirecting you to the right FAQ on the website immediately which provides you with the best and fastest response possible. This can be predicted by matching on the context of the question to the expertise of customer service employee or the particular web page. This will exceed the customer’s expectations because of the quick and efficient calls while the insurer enjoys less costs by saving time on helping every individual.
A third way to improve customer experience is to recommend the right care providers. This can be done on a personal level – a young adult male might have a better connection with an adult male than he has with an older female – or on a professional level, i.e. specialism or effectivity. If the claim history data of the insurance provider indicates that care provider A resolves a certain medical issue in on average five sessions and care provider B resolves it in three sessions, provider B is predicted to be the recommendation for the best revalidation. This results in less costs for the insurance provider, because of less treatment to be claimed and the best possible care for the customer resulting in fast recovery.
Another crucial factor in customer experience in insurance is a quick and correct payout of the claims. But as insurer you might know that not all claims are valid, thus claims have to be reviewed. Unjustified claims (or to put it bluntly: fraud) cost insurance companies around 80 billion euro a year (source: Coalition Against Insurance Fraud). Preventing fraudulent claims from being paid would result in a significant decrease in costs. But reviewing every single claim is an expensive and time intensive task. Data Science assesses the probability of claims being fraudulent, which enables the insurer to only pass through high risk claims to employees for inspection. This results in a more efficient claim inspection by only putting effort in high risk claims. Hence, the ROI on invested time is expected to improve significantly. This solution will also enhance the number of satisfied customers, because the claims that have a low probability of being fraudulent will be paid out immediately. Can you imagine how happy one could be when his phone has just been stolen, and he receives the payout immediately? Sure he will be, because it’s partly your service as insurer that got him in touch with his friends again so quickly!
Fraud detection with Data Science
Insurance claim prevention is all about putting the customer journey first. Generally, consumers take out insurances because of the fear something bad might happen to them. Overall, this is a pretty negative framing for a product proposition. One could put it more positively if insurance providers would not provide financial aid to customers when the damage has already occured , but also provide valuable advice to avoid these unpleasent real-life experiences from happening.
An example can be found in health insurance. Hospitals and insurance providers are inventing ways to use Data Science to detect patterns and causality in the data they possess. This makes diagnosticating in an earlier stage of the illness possible, when treatment is more effective and less intense. For instance, when a policyholder is identified with a highrisk of getting diabetes, based on behavioral data from the policyholder and his peers, he can have a free consultation with a company doctor. The doctor can provide some valuable advice on how to lower the risk of getting diabetes.
The possibilities of Data Science in healthcare
Interested in this topic? Read more about it in our article on insurance claim prevention.
SHOULD WE FORGET ABOUT PREMIUM?
Not at all! We should remain focus on premium differentiation too. Since insurers possess vast amounts of data which give data scientists the ability to assess factors as risk and loyalty on an individual level. These predictive insights can be used for individual premium setting. Hereby it is possible to offer competing coverages to your customers by adding predictions on the pricing strategies of your competitors and taking these into account to offer the best possible offer to an individual.
For example, by using external data from RDW, we concluded that owners of turbo powered cars drive faster and even more recklessly than owners of a non-turbo powered car. Therefore, these drivers have a higher individual risk and are expected to file higher and/or more claims. This information serves as important input for premium setting on an individual level. Read more about this in our use case.
To conclude, insurance providers want to gain and keep a competitive edge, therefore innovation is key. Risk assessment and premiums will always be important but become a commodity, so a shift to a more customer centric view is necessary to stay ahead of your competition. Reviewing the entire customer journey, other parts of an insurers service will become more important. Luckily, increasing amounts of data are accessible to insurance providers, and Data Science is here to turn data into value. Let’s turn these ideas into action!