Artificial intelligence is having a positive impact on the industry. There are substantial use cases in almost every part of the insurance business, from policy administration and claims to customer service.

AI can improve fraud detection and claims processing without human intervention from reporting through capturing damage, audit and communication with the customer. For customer service, AI-based chatbots can assist with questions that traditionally required a call center or agent.

A 2017 Accenture report even found that 74 percent of insurance customers would be willing to receive computer-generated insurance advice about coverage options and recommendations.

Use cases for AI in underwriting are also promising but more challenging considering the regulatory environment. To date, machine learning has been used to develop predictive models that are black boxes and only show decision output. This is at odds with regulators that want to know what variables are influencing the decisions in order to avoid price optimization concerns. The technology has been scrutinized dating back to 2014, perhaps most notably surrounding Earnix’s growth in the area.

While AI will certainly gain more traction in insurance over the next year, what’s most material for driving the insurance industry forward today is the maturity and evolution of data analytics. Despite significant advancements over the last decade, insurers are still having a hard time turning burgeoning data into actionable business decisions and intelligence. It’s only been a short time that the chief data officer, who focuses on organization-wide data strategies that consider ethics and data protection concerns, became prevalent in the industry.

Addressing the opportunities for new data and technologies within underwriting, for example, is typically an added responsibility for the chief underwriter or actuary—and neither of those individuals may be closely connected to the business side of the organization. This structure can slow new data initiatives as those positions are responsible for convincing the C-suite about the benefits.

It’s important that insurers become more agile and informed going forward with leaders that represent both IT and business interests.

Using data to address organizational needs isn’t just the purview of the insurer but also of the technology providers that assist in all areas of digital transformation. At this year’s InsureTech Connect, one of the leading insurance technology trade shows, there was a gap between startups pitching technology vs. vendor solutions supporting a true business case. True progress will stem from technology maturity of both organizations and their tech providers.

The 2020 race for larger shares of the small commercial market will singularly drive more sophisticated use of predictive analytics to streamline the quote-to-bind processes. Small commercial policies are a particularly challenging segment. First, these businesses’ purchasing behavior is more like consumers. Despite the complexity of commercial insurance, customers expect an experience like shopping on Amazon. The second challenge is on the agent side. Agents are incentivized to focus time and resources on larger policies that generate more commission, pointing to the need for more automation in underwriting that can limit human intervention to quote and bind small business policies.

Both challenges indicate a requirement for better straight-through processing (STP). Small commercial is also responsible for an increase of MGAs that are leveraging digital platforms and advanced analytics to deliver quotes in under a minute.

Insurers will require more granular third-party data to accurately price small commercial policies. Many large insurers simply don’t have enough in-house transactional information on these small accounts, creating blind spots in their datasets that make it challenging to build accurate predictive models. Transactional data includes historical and updated details on individual policy and claims information, such as changes in exposure by class, drivers and premiums pre- and post-audit, over several policy terms. Insurers likely receive similar broad data insights from bureaus such as ISO and NCCI, but sources based on statutory filings provide a more aggregated view, as opposed to updated snapshots of risk throughout the year. Insurers must look to other data consortia that have transaction-level data to fill in the blind spots with small commercial policies. We’ve worked with dozens of insurers to leverage these data points for writing business in new geographies or simply to better assess risk in their regions.

Not doing so places incumbent insurers at risk of losing their competitive edge to InsurTechs focused on this market that are unencumbered by IT backlog and other institutional barriers.

Insurers must rely on third-party sources that possess this type of detailed information and combine the right in-house and external datasets to develop variables that are more predictive for their specific needs. This is easier said than done, considering how expensive and time-consuming the process is for resource-strapped insurance companies that have many IT priorities. Focusing on a more efficient way to process this information should be a major focus in the coming year.

As insurers continue to address analytics at a more strategic level, they will also require more precise reporting that monitors and measures the progress of analytics initiatives on their book of business. The tools must look beyond an aggregate view of their portfolio and drill down to details by specific hazard group (if workers compensation), geography, or even the performance of underwriters and agents at the individual policy-level. Identifying specific patterns and trends will allow insurers to better assess the impact of their analytics programs ahead of time and make the necessary adjustments without dismantling areas that might otherwise be effective. It’s best to rely on advanced data and analytics providers that have a deep understanding of the insurance industry and how predictive model decisions affect each segment of the business.

As the insurance industry enters a new decade, it will be a new battleground for predictive analytics. The last 10 years largely embodied the division between “have” and “have-nots” of using analytics to improve risk assessment and pricing, and the next decade will focus on which insurers can better connect advanced analytics with business strategy. Those that can synchronize data initiatives with overall business goals will continue to be competitive as the market continues to change.