In a prior blog post on aligning predictive model development with the product development lifecyle, I shared that, when developing a predictive model, a large amount of data is required. The point is, commercial lines carriers need sufficient performance data to train and validate whether their model effectively predicts their target in the first place. Regardless of how talented their analytics team is, the data they leverage must be credible. However, there is another element of successful predictive modeling that has to do with organizational style.

When leveraging predictive modeling, commercial carriers typically fall into one of the quadrants illustrated in the matrix below, depending on two key factors:

  1. The degree to which the culture supports analytics
  2. The level of access to credible data

Predictive_Modeling_Culture_Graphic

I: Data-Informed

A data-informed carrier has both sufficient credible data and a supportive culture—what I call “the sweet spot”. However, for carriers in this position, their work is NOT over. The efficacy of the data and the talented experts that leverage it should be monitored on an ongoing basis to ensure the models continue to perform as intended. If a model is no longer delivering the desired results, it’s time for a deeper dive into the underlying causes. For example, was there a cultural shift or does the model need to be recalibrated to reflect changing circumstances?

II: Data-Driven

Is it possible to invest enough in credible data sources but have it miss its full potential due to a lack of culture that supports it? Absolutely! A solely data-driven carrier’s culture simply won’t embrace the power of predictive analytics, regardless of the available data. Successful predictive modeling depends on having a team that’s not only capable of but also willing to leverage the insights offered by their data for improved decision making.

III: Potentially-Lost

“Potentially-lost” carriers lack both sufficient data and the right culture to support predictive modeling. This is the least favorable position within the matrix. There may be a variety of reasons an organization finds itself in this position. Regardless of reason, the mission should be to progress by any means possible to improve overall performance. According to Eric Siegel, founder of Predictive Analytics World and executive editor of The Predictive Analytics Times, “Predicting is better than pure guesswork, even if not accurately, delivers real value. A hazy view of what’s to come outperforms complete darkness by a landslide. The Prediction Effect: A little prediction goes a long way.” Any effort is better than no effort.

IV: Culture-Driven

Some carriers have a supportive culture ready to embrace predictive modeling, but lack quality data. However, a culture-driven commercial lines carrier is in the strongest position for easy improvement and often leverage a third party to take their predictive capabilities to the next level. When leveraging a model for underwriting and pricing, these types of carriers consistently use their model to inform their decisions. An underwriter may choose to ignore the results of the model, but only after careful deliberating and documenting why. Some of these types of carriers are challenging the status quo by engaging with a partner like LexisNexis Risk Solutions to build the next generation of analytical solutions by “pooling” their data with other carriers to leverage best-in-class analytical models.

Where do you fit?

Data and culture should not be competing forces in embracing predictive modeling. It takes the right mix of both to get the best results. A good balance of credible data sources and a culture open to leveraging predictive modeling is the key to garnering actionable insights that lead to more accurate risk assessment and pricing based on loss propensity.

For additional information about predictive modeling best practices, please see our whitepaper, Making Predictive Modeling Work for Small Commercial Insurance Risk Assessment.

Please contact us directly to find out how our predictive modeling can become part of your standard workflow to help you evaluate a business’s loss propensity at the time of quote, underwriting or renewal.

Post a Comment

Your email address will not be published. Required fields are marked *