Can predictive analytics forecast if predictive analytics will transform the insurance industry? 

If you were to ask someone ten years ago, you might have thought ‘yeah right.’

But in the last few years, insurers have been investing in data and analytics, and in technology and infrastructure. Their aim is to capture and store more data than ever before, and to build talented data science teams that can turn data assets into valuable, and actionable, business information. There has been a constant stream of adverts for senior data science roles, all tasked with finding how machine learning could add value to insurance businesses. A great insight into how data science can unlock value for insurers can be seen in an interview with Aviva’s Global Director of Customer Analytics and Data Science, Orlando Machado.

So, what should we expect from all of this? The answer seems to be that investment in predictive analytics (and more recently machine learning), will improve customer experience significantly, whilst cutting claims handling time and costs, and eliminating fraud.

Insurers will be able to fast track claims, and process them with little to no human intervention. This is already a reality for many companies like We Predict which uses predictive analytics to enable vehicle manufacturers and suppliers to manage the frequency and cost of malfunctions for vehicles under warranty. US insurer Esurance has taken to using predictive analytics as a means to skipping adjuster inspections on motor claims related to Hurricane Harvey.  

Hurdles and barriers along the way

Integrating predictive analytics with business processes will not all be plain sailing, however.

Spotting troublesome claims early is a bit harder; figuring out strategies to mitigate the risk once identified is trickier still. Bridging both the technology gap and the communication gap between a data science team and a claims organization takes effort.

To start with, the information needs to be delivered in a timely fashion (preferably seamlessly), into the workflow of the adjuster, and with a notification to the supervisor. In addition, the information delivered needs to not just raise a red flag, but to tell a story about the risk identified, and propose a solution.

It takes more raw data, and more observations, to spot large claims than it does to settle small claims on an automated basis, simply because small claims make up the bulk of observations in any given data set. It requires even more data to build an artificial intelligence (AI) capable of handling the complexities of the largest claims. The inherent complexity of each large claim tends to mean they are more individual than the high volume, low value ones that can be fast tracked. 

Overcoming this challenge

To solve this challenge insurers are turning increasingly to external data sources; adding more information about a claimant or injured party, such as identity verification or social media data. The purpose of this is to paint a more complete picture of a claimant, which may assist, speed up, and reduce costs in claims processing. As it stands, there are limits, even outright barriers, to just adding external data points to an internal pool of claims. However, new data from technologies such as Internet of Things (IOT), that the insurer may own, or have easier access to, may well change this.

Machine learning requires a lot of data. Even deep-learning techniques are not great at dealing with, and understanding, patterns they have never seen before. While the insurance industry is data-rich in many ways, when it comes to the sheer number of events that are the bedrock of other industries, the insurance claim is nowhere near as frequent. Google handles nine billion searches per day. Visa has 300 million transactions. Amazon receives almost 200 million visits per month. Insurers only deal with a tiny fraction of that volume for claims, but insurance businesses stand to benefit by using predictive analytics to improve the efficiency and value of their systems. If data gleaned from external sources can prevent or reduce claims, this may produce significant cost savings, critical for all insurers.

Predictive analytics in the age of machine learning offers insurers the opportunity to improve processes on a scale that was previously impossible. What is not to love about access to easily consumable information, that will transform quality of insight and customer service? With the ability to enable smarter insurance operations, reduce costs, and grow profitability, this a no-brainer. Predictive analytics is clearly set to revolutionize the insurance industry.

Follow me on Twitter here.