The Need for Transparency in Underwriting

Open the black box and combine analytics with underwriter expertise to evaluate the computer’s conclusions and where the information comes from.

Computer keyboard

The commercial underwriting game has changed. Data analytics, artificial intelligence and machine learning provide access to more information from a variety of sources, including social media and publicly available databases. But while many technology solutions have been quick to provide answers to underwriters, they’ve been less focused on offering information about how and what data is used to provide conclusions. “The answers are in the algorithm” has been a common explanation about the ways the information is applied, sourced and weighted to answer insurance questions. 

But what happens when you open that black box and combine analytics with underwriter expertise to evaluate the computer’s conclusions and where the information comes from? The answer: You get more accurate, efficient underwriting, better customer service and more business. 

What does transparency look like?

Transparency doesn’t mean sharing the secret sauce or giving everyone access to intellectual property. It does mean pulling back the curtain so those using the solution can easily see where the information is coming from and make decisions with that information in mind. 

Imagine this scenario. A restaurant is looking for insurance. In the application, the restaurant said it does not do deliveries. To speed up the process, the underwriter working on the file uses an AI-enabled solution to comb the internet for information and automatically pull in answers for the underwriting questions. The solution says the restaurant does, in fact, do deliveries. Because the underwriter is strapped for time, they aren’t able to do the manual work to figure out why there is a discrepancy. The underwriter accepts what the solution provided and denies the coverage because that carrier does not insure delivery restaurants. This leads to a long back and forth with the agent and customer to figure out why the coverage was denied, as the customer is adamant that it does not deliver. Not only would the carrier lose a potential client, but it also diminishes the carrier’s relationship with the agent. 

Now consider the same scenario. The only difference is that the underwriter is using a technology solution that’s transparent. The agent can see that there is a discrepancy between what the agent submitted and how the solution answered the question. The underwriter can go directly to the sources where the delivery information was pulled to see that delivery is via a third party, such as Uber Eats. Because the restaurant doesn’t do deliveries itself, its insurance category doesn’t change, and the carrier can write the risk. The underwriter is able to approve the policy without time-consuming work and a long delay between the agent and the customer. 

See also: Eliminating AI Bias in Insurance

Machines don’t replace humans

The value of human expertise is fundamental to the success of the sector. So why are some solutions bypassing professionals’ insurance acumen? Commercial insurance is complicated. There is no one-size-fits-all approach as businesses each have their own particular sets of risks. While AI and data analytics platforms drive underwriter efficiency, it is equally important to provide all of the information they need to easily identify and resolve unique circumstances and make sure customers get the right coverage. 

Advocating for transparency in analytics

Carriers should advocate for transparency. It can save their underwriters significant time and help fuel  business. Here are three reasons why data transparency is critical to insurance buying: 

It builds customer trust. “That is what our underwriting program determined,” is not going to suffice if a customer is wondering why they were denied coverage or their premium ended up being significantly higher than what was quoted. Customers understand businesses use algorithms to speed processes, but they also know the algorithms can be inaccurate. Being able to tell customers exactly where the information comes from and how it is used can help underwriters give clients clear answers to their questions. 

The human-computer combination enables faster interactions. Carriers get the best of both worlds: computer speed coupled with underwriter expertise. Underwriters can make decisions faster. If there is a discrepancy, they can easily resolve it by checking the sources for themselves, eliminating the need for manual search. Finally, underwriters have confidence that the information they are using is accurate. In a non-transparent platform, if an underwriter determines the solution answered a question incorrectly, it could lead them to wonder what else the solution might have gotten wrong. 

It keeps you ahead of the data compliance curve: Consider California’s Consumer Privacy Act or Connecticut’s Personal Data Privacy and Online Monitoring. More and more states are enacting data privacy legislation, while the federal government is also working on passing legislation on its own. Data privacy issues and unintended bias in data analysis are growing concerns in the sector. Are solutions using personal information about a company’s leaders or employees that could infringe on their privacy rights? Are algorithms delivering different results for different demographics of people? Carriers that can show exactly what information was used to underwrite a business can easily appease any regulatory concerns. If there is an issue raised about the fairness of price, for example, the carrier can easily pinpoint the text used to determine the premium and show a regulator that there was no bias or overreaching into employee personal information. 

Data analytics and AI are increasingly becoming table stakes in insurance underwriting. Now the conversation needs to move to transparency. Easily accessing the data sources that solutions use to determine their answers puts the final decision-making into the hands of insurance professionals – where it belongs. Combining technology efficiently with human expertise enables a more efficient and accurate underwriting process to better serve customers and ultimately grow business. 

Prakash Vasant

Profile picture for user PrakashVasant

Prakash Vasant

Prakash Vasant is co-founder and CEO of NeuralMetrics, which provides fast, transparent, accurate and actionable commercial underwriting answers and intelligence for carriers and agencies. The platform leverages AI and natural language processing to analyze unstructured and public data to automate and improve the underwriting process.

He is a serial entrepreneur, launching and scaling successful ventures in finance and technology and leading global teams. Prior to NeuralMetrics, Vasant led a global IT consultancy as well as served in senior positions with an inter-bank currency dealer and a corporate foreign exchange advisory consultancy.  

Read More