Big data and predictive analytics have not yet altered the course of the insurance industry, but they will. Over the next decade, the use of advanced analytics will have a dramatic impact on the insurance space in three key areas. First, I predict big data will contribute to reduced friction in the relationship between sales agents and potential customers. Second, I believe machine learning will ultimately replace human predictive analytics. And third, I predict that investment in the data industry will itself drive change in the insurance space. According to a report from 451 Research, an information technology research and advisory company, spending on the overall data environment across all industries will grow from $70 billion in 2015 to $120 billion by 2020.
How Big Data Will Reduce Friction in the Insurance Industry
Insurance is generally something that is sold and not sought, and insurance companies spend a significant amount on selling insurance. It has been this way for decades. However, through the use of big data and advanced analytics, insurers have the opportunity to simplify the purchasing process, making it easier, faster and simpler for consumers to buy and keep insurance. Advanced analytics will impact:
"The key to success over the long term is to begin using machine learning in the insurance industry"
• Sales and Marketing. Finding people who are at the right point in their lives to purchase the right insurance product is probably the most difficult part of the process. For every sale they make, insurance agents have to talk to a number of potential customers who may not be in the market to purchase a product. The sometimes high cost to market and generate leads can contribute to the difficulty for talented insurance professionals to find the right buyers. Enter the promise of big data. Big data has the potential to make it easier for insurance professionals to connect with interested buyers. By combining internal carrier sources of data with external providers of data, including social media, insurers can better equip agents to target potential customers.
• Underwriting. Underwriting is another excellent example of how big data can help reduce friction in the insurance industry. By using big data to source and analyze prescription and medical data sources, insurers can reduce the need for certain medical requirements in the underwriting process, especially for lower face value policies.
• Fraud. The insurance industry is a frequent victim of both provider and policyholder fraud, which leads to higher costs for everyone. Through the use of big data and predictive analytics, insurance risk professionals can help business users better detect suspicious activity and thereby eliminate costs associated with insurance fraud.
Machine Learning Will Take Over for Human Predictive Analytics
The rate of data growth today is extraordinary. An EMC study predicts that the digital universe will grow from 4.4 trillion gigabytes in 2013 to 44 trillion by 2020. How is the insurance industry handling this rapid growth in data? Though it has been an information industry since it began, many insurance carriers are not fully leveraging the data they have today in their systems.
Why is this? Think about the industry’s origins. The first mutual insurer, Equitable Life Assurance Society, created age-based premiums based on mortality rate in 1762 in the United Kingdom. While the actuarial fields in insurance have used data since the beginning, the industry has not consistently used the data it has to drive sales and improve the customer experience.
Meanwhile, other industries have flourished using predictive analytics to help drive sales and customer experience. Netflix, for example, uses machine learning algorithms that have allowed the company to extend its services to more than 190 countries.
The key to success over the long term is to begin using machine learning in the insurance industry. Too much data will come from IoT devices, wearables and social media for human data scientists and actuaries to keep up. The usage curve will be exponential rather than linear.
Breaking Through with Current Big Data Investments
How do we make existing big data or new big data investments successful today? I believe there are key steps in both the short term and the long term for these programs to be successful.
• Start Small—Manage your scope down. Start with one or two predictive models, and scope out the tests in days rather than weeks.
• Bias for Action—Focus on starting and completing your models. Put your campaign to work. Don’t get caught overanalyzing the model. Let your results prove the effectiveness of your campaign.
• Data Governance—Insurance companies have data spread across the enterprise. Getting data governance in place to standardize data dictionaries and coordinate the movement and management of data through the company will allow your organization to move forward with greater efficiency. This is not a requirement to start, but over the long term you will need this in place to be successful.
• Data Infrastructure—Over time, modernizing your data infrastructure will lower costs and improve results. Tools like Hadoop and data lakes, which help leverage unstructured data throughout the enterprise, can help organizations implement modernization with greater efficiency. While it is not necessary to start analytics with a modern data infrastructure, over time it will improve results.
• Machine Learning—Machine learning will be the path for predictive analytics for the insurance industry. It will take time but it will have a dramatic impact once implemented. Take time to plan and determine how machine learning can augment your data scientists.
The use of big data and advance analytics is already impacting several industries on a large scale. Industries are being unbundled and re-bundled as part of the overall digital transformation. The insurance industry will be no exception. We can pave the way for big data projects by starting small with a bias for action. Bring in data governance as soon as possible. Layer in machine learning and modernize your data infrastructure to bring additional results.