Applying data science strategies and advanced technologies to the underwriting process starting with the submission process can help underwriters identify and write better business and ultimately improve profitability.
Two Sigma’s Barry Liang, vice president of product management, and John Paladino, senior vice president and head of early partner development, reveal how carriers can take incremental steps to relieve the burden of manual and unsystematic work and become a more efficient, effective, data-led organisation.
Where are insurers missing critical data?
John Paladino: The submission process historically has been quite manual, cumbersome and typically includes lots of handoffs. A lot of data doesn’t get captured because of all of the friction within the process. Valuable data are often squandered during the submission process. For example, most carriers only maintain records of quoted and bound policies for underwriting and policy servicing purposes.
However, there is tremendous value in the ‘submitted and not-quoted’ application data, which is not being captured today. Having a systematic way to collect this information can help unlock insights into the underwriting process and/or support ad hoc analytics use cases.
Barry Liang: A lot of the behavioural data and the activity that takes place during the underwriting process isn’t captured, for example a carrier often doesn’t know the reasons why an underwriter changes a policy.
Did they perhaps dial limits up or down during their broker negotiation for some reason? Having a system that allows you to capture that level of decision-making can help you map out your decision tree and make more informed decisions in the future.
There also needs to be a more proactive approach to the use of third-party data. Underwriters may search for specific datasets to validate an assumption, but what is really needed may be insights derived from another type of dataset. It is important to think through the problem you are trying to solve and to strategize as to the type of information that will help make the most effective business decisions.
Finally, we often hear from our partners and clients that it’s too costly to do advanced analytics projects because of all of the data-cleansing and preparation that is needed beforehand. An insurer’s data isn’t always in the proper format or isn’t organized in a way that allows it to be utilized properly in models.
If carriers are managing on current systems, what is the motivation for change?
Paladino: One of the motivations for change is the need and desire to keep up with innovative competitors. Many of the more innovative carriers have seen the technology transformation and successes other industries have experienced in applying data science and machine learning applications.
While change is good, it does need to be managed properly to minimise execution risk. Through our conversations with insurance companies, many view taking on large transformational change as hugely disruptive. Ripping out and replacing old systems all at once is often a multi-year project.
Carriers are more apt to innovate if transformative change is incremental and of relatively low risk. This is why we have chosen to tackle the issue in smaller, digestible pieces. A focus of delivering immediate value to the end users encourages adoption. That’s why we have chosen to start with the submission process.
There are of course, many challenges insurers need to address but most view their submission process is currently suboptimal and suffers from a large data gap.
What are the overall benefits, beyond cleaner data, that might drive carriers to look to data science as a solution?
Paladino: One thing we’ve seen as a struggle for the industry is its ability to leverage third-party data. Carriers first must think through how they are going to use data. The application must make sense for the business. We often hear about carriers using third-party data to streamline operations, but integrating third-party data into decision-making and the evaluation of risk is where the magic begins.
We have been using data and data science to help carriers prioritise submissions so that underwriters can focus on the best business opportunities. We filter out much of the noise in the submission process by bringing forth the best business opportunities and reduce the need for excessive human intervention.
Liang: In the end, policyholders benefit. They can buy faster from the carrier, enjoy a more holistic set of coverages because data insights are available to help carriers identify and be more informed around risk exposures. Policies are more appropriately priced and brokers can interact with clients and carriers much more seamlessly.
