Unlocking growth with speedy underwriting submissions clearance

After working with this client for more than 15 years, we'd already eliminated non-value-added activities across underwriting, claims, and technical accounting. But the company, keen to explore new digital technologies, wanted more.

We looked for opportunities to reimagine how processes are run and managed. Since operational efficiency was a core outcome, we focused on areas with the scale for potential automation.

We zeroed in on the underwriting submissions managed by a front-desk team of about 60 staff. A design-thinking workshop with all those involved in the process allowed us to reimagine and redesign submission management to allow for automation and improve the end user experience.

The front-desk team dealt with submissions for new policies, renewals, and endorsement requests. All these emails contained vast amounts of unstructured data in attachments, such as Excel, Word, and PDF files. This was the biggest challenge we needed to solve – taking away the pain of having to review the content of every email, find the relevant data elements, apply business rules and validations, and then send the submission to an underwriter for review. A task made more difficult by switching between two screens – extracting data from one and keying it into a different system on the other.

This was all time-consuming, error-prone, and tedious.

Although robotic process automation (RPA) is one technology that can provide the efficiency and ROI the client was searching for, we knew it wouldn't come into play in this project. RPA works best with structured data, but the submission data came from brokers in different formats, with changes in the position of the data and types of content. So more intelligent technology was required – one that could leverage artificial intelligence to read emails and attachments and extract and structure the relevant information.

We put our minds to work to create the ideal solution.

To bring structure to submission data, we built Cora Front Desk Assistant, an AI solution that reads and understands natural language, document formats, context, and document structure details. The engine autonomously builds a deep understanding of this material and the data in it – making it ideal for processing free-form text, tables, and hard-to-recognize data fields. Even better, it keeps learning over time to improve accuracy.

By applying natural language processing and natural language understanding capabilities, Cora Front Desk Assistant can read emails, extract knowledge from the email body and attachments, and run compliance checks or other business rules to provide end-to-end automation of the submission clearance process.

The digital investment paid off. In fact, it has already far exceeded the results the client expected to see. Among the business benefits:

• A trebling of straight-through acceptance rates
• The time taken for customer acknowledgement plummeting from between two to five days to four hours
• Up to 50 percent quicker time-to-revenue through streamlined processes
• An increase in Cora Front Desk Assistant accuracy from 53 percent to 85 percent, as it has learnt from the submissions it's processed over the 10 months it's been operating
• A 40-50 percent reduction in the average handling time
• A reduction in the total cost of handling submissions of 30 percent – a projected saving $3 million over the next five years.

The bottom line: customer turnaround times are shorter and require fewer touch points. Plus, improvements in accuracy, user experience, and resource efficiency mean the front-desk team can now focus on more value-added work.