In 2017, 62% of insurers incorporated predictive analytics into their underwriting, according to the "2018 Outlook Report" from Valen Analytics. The majority of insurers are utilizing data-driven approaches to better understand their risks in underwriting commercial business.

But the business advantage of analytics enjoyed by early adopters can mistakenly be viewed as less valuable today, as more companies catch up.

Insurance companies with "first-mover advantage" must now tighten their underwriting practices and guidelines around predictive analytics, taking a fresh look at their models to retain and enhance their competitive position.

Here are some guidelines to help insurers fine-tune predictive analytics to improve ROI and increase profitability.

Related: Fully leveraging data analytics in insurance

Perform a model evaluation every year

Each predictive model and use case differs, but it's advisable to assess the model every year to monitor performance. A simple baseline indicator of the model effectiveness is how the lift, or the risk assessment accuracy, has performed year-over-year. If the lift is declining, it's important to identify why. There are many factors that could lead to a model under-producing over time including changing rates, market conditions, or a shift in the insurer's strategy.

Models often need changes to update rates or LCMs, and can be recalibrated to boost performance given other considerations, like growing into new markets. Whether it's simply maintenance or executive strategy insight, incorporating these measures will best optimize the model.

Adjust underwriting modeling guidelines as needed

One of the largest contributors to predictive modeling success is how insurers balance data and underwriter experience. In Valen's example, underwriters have the ability to provide feedback, based on the results of the model. This functionality allows for management to consider a full scope of both model scores and underwriters using their experience to make the most informed decisions.

Valen clients find it effective to send specific policies through straight-through processing (STP), such as small premium or low-risk policies. It frees up underwriters to review higher risk policies and apply their expertise where it matters most. Commercial lines insurers also balance data and expertise by having certain parameters based on the model's recommendation for crediting or debiting accounts.

When implementing a predictive model, it's important to monitor its use and real-time results to maintain consistent lift and ensure results don't deteriorate. Just like any integral process, this powerful decision-making tool needs a solid implementation plan that includes regular upkeep.

Dax Craig is co-founder and CEO of Valen Analytics. He can be reached by sending email to [email protected].

The opinions expressed here are the writer's own.

Read more columns by Dax Craig:

From model ready to business ready: Connecting data to ROI

Variety and context are critical for predictive analytics success

A proactive approach to working with regulators