Insurance is one of three major industries that will be transformed within the next decade by big data, artificial intelligence (AI) and machine learning, notes Bernard Marr, Forbes contributor and author of “Big Data: Using SMART Big Data, Analytics and Metrics to Make Better Decisions.” “The more data the insurance industry has access to, the more they can automate, and the more they rely on AI, the more the industry will change,” he says.

For many insurers, the concept of applying elements of AI or machine learning to their enterprise strategy may seem overwhelming. But the fact is, actuaries and insurance statisticians have used historical data for years to recognize patterns in claims and underwriting that are tied to predictive analytics.

Fortunately, the next iteration of this technology is upon us and already accessible to the property & casualty insurers for use in a variety of ways, from analyzing natural language patterns to analyzing policy language, and to analyzing naturally occurring trends in a book of business to improve risk assessment, pricing and underwriting. In the claims area, machine learning can assist with fraud mitigation, and with telematics, with auto loss prevention.

In McKinsey’s “An Executive’s Guide to Machine Learning” report, authors Dorian Pyle and Cristina San Jose say that there is a more urgent need now to embrace the prediction stage.

“Today’s cutting-edge technology already allows businesses not only to look at their historical data but also to predict behavior or outcomes in the future—for example, by helping credit-risk officers at banks to assess which customers are most likely to default, or by enabling telcos to anticipate which customers are especially prone to “churn” in the near term.”

McKinsey views machine learning as part of a larger corporate strategy: C-level executives will best exploit machine learning if they see it as a tool to craft and execute a data strategy vision. “Without strategy as a starting point, machine learning risks becoming a tool buried inside a company’s routine operations,” say Pyle and San Jose.

Putting that strategy together requires some specific execution steps, notes Barry Ralston, AVP of Business Intelligence at ISCS, a developer of SurePower Insights™, a business intelligence offering for property and casualty (P&C) insurers. “Identify the (business) problem you are trying to solve, and begin with an end in mind,” Ralston says. “This means a clear, strategic definition of what your ‘success’ looks like, with the caveat that your data may point your organization in another direction.”

Ralston says it’s also important to identify the data required for a machine learning exercise. “All other things equal, target a business problem for which real-time data is readily available. Likewise, favor an area of investigation for which multiple facets of the problem are represented in related, yet separate, data sets. More data beats a better algorithm,” he says.

Once the data is identified, it must be enriched. “Resist the temptation to allow your machine learning platform to process data in a raw form,” cautions Ralston, who provides the example of an attribute as simple as a transaction date. “Ensure your machine learning platform has the full picture of the data element,” he says.

Finally, Ralston says, it’s important to start small, and apply machine learning models iteratively: “As you create models for your specific problem, apply a process to evaluate the quality and expected performance of each solution.”

The McKinsey authors agree with the “start small” advice, but for additional reasons tied to the larger corporate strategy: “Look for low-hanging fruit and trumpet early success,” say Pyle and San Jose. “This will help recruit grassroots support and reinforce the changes in individual behavior and the employee buy-in that ultimately determine whether an organization can apply machine learning effectively.”

Ralston says ISCS sees the advantages of insurers leveraging elements of the algorithmic approach of machine learning and the self-learning element of artificial intelligence. “We are creating more intelligent software solutions that actually improve how they function over time,” Ralston points out. “This ensures that SurePower Insights’ analytic and business intelligence capabilities enable customers to access and understand critical core insurance data, and therefore, make better-informed business decisions.”

Related Webinar: Machine Learning for Insurers? Start with Business Intelligence