There is a concept in economic and social sciences called “noise” garnering a lot of attention lately. Simply put, it’s the idea that given the exact same set of facts, two equally qualified people will not make the same decision. It can be proven again and again across professions, and unlike “bias” in which people make decisions based on a deeply held opinion or cognitive logic errors, noise is completely random. Under scientific conditions, you can show that the time of day, the weather, if the decision-maker’s team won last night, and many other seemingly random variables all factor into how every human disposes of decisions. We like to think we’re fair, impartial, and objective in making our decisions, but evidence doesn’t support that.
Why does “noise” matter in insurance? If your underwriter is well-rested, fresh on a sunny morning at 9:00 AM, and just paid all of their personal bills last night with plenty of money to spare, he or she may be so upbeat & optimistic that they err on the side of the insured, forgiving a little red flag here and there. If your underwriter is hangry at 2 PM, it’s cloudy and rainy out, and they just got a late payment notice from the utility company, they are more likely to be harsh in their judgment. So, both are “noise,” (rather than a consistent bias) that swing the pendulum in opposite directions, but neither is accurate, and both pose risk to your business.
Where else is there noise in insurance? Well, how many non-automated processes do you have? Every time there is a human decision made in your business, noise comes into play. Which adjuster do we use? Which vendor? How much do we offer for settlement? Do we take it to court or settle? Seems like a lot, but there is an answer.
How do we combat noise? Algorithms. Given the surge in InsurTech modernization happening across the board, you’re probably already using them in some areas of your business today. Claim severity & complexity scorings are a common one, but if you have a human review process following the algorithm’s decision, you just re-introduced the exact same noise that the algorithm was designed to prevent. The better approach is to invest in a self-learning, AI-enabled algorithm that will improve and consistently reduce noise over time, whereas the humans will not.
When you couple an algorithm’s capacity for fast and consistent decision-making with the consumer demand for real-time self-service in today’s market, you see where the trajectory points. Carriers who are both accurate and fast are positioned to ride the tide of demand for speed without undue risk to their business. Those that cannot should expect turbulence ahead. Every insurance company should be actively evaluating potential noise in the business… where is yours?
More reading: Noise: A Flaw in Human Judgment, by Daniel Kahnema
