Policy Checking

RiskGenius took on the challenge of policy checking for the commercial insurance industry. It's tedious and often skipped, which leads to potential legal exposures. I designed a solution that reduced policy checking completion times by 70%.

6 min read · InsurTech · User Research · Product Strategy · Prototyping · Interaction Design · Visual Design

Some insurance context

Checking the contents of policies is one of the simplest ways for the insurance industry to prevent massive error and omission lawsuits. Training new associates to perform accurate checks can take up to six months, and the manual comparison of policy documents is tedious and error-prone, so it's often just skipped. This results in no one knowing what is actually in a policy until a claim is filed and legal experts are involved. We found this same situation across the industry.

The declarations pages from three insurance carriers (above) show the significant differences in how policy information is commonly presented. All names and policy information have been anonymized to protect privacy.

The initial strategy

To address this challenge, we initially focused on parsing the values from the declarations page (the executive summary of key policy details) into a table. Using the extracted information, analysts could compare policy documents and conduct thorough coverage evaluations. Accommodating the diverse formats used across the industry posed a significant technical hurdle for our data science and engineering teams, so we needed a better understanding of the policy checking process and problem space in general. 

The first version of the policy checking solution extracted the "messy" information from the policy forms into a clean new table with filters and sorting features. The idea was to pull everything into one place for a quick overview.

Digging much deeper

The primary research team consisted of the Head of Product and myself as the sole Product Designer. We soon recognized that extracting information into a new table detached the values from their context, which is vital for the policy checking process. Additionally, we discovered that brokers and carriers relied on checklists to ensure quality and considered them to be some of the "secret sauce" in their service offering. These tools were typically complex spreadsheets or simplistic text documents.

I realized that checklists held the key to our software solution at the same time our engineering team was recognizing the need to rebuild our backend infrastructure. We needed to support the substantial volume of customer documents being uploaded and processed to train our indexing algorithms. Seeing this as a unique opportunity, I successfully made the case to leverage the rebuild to pivot the product and create a checklist-driven solution.

Paper checklist used to analyze custom manuscripted policies
During development of the new checklist solution (and the simultaneous rebuild of the backend infrastructure), we had several collaborative working sessions and design sprints that brought the entire distributed team together.

A new approach

Collaborating closely with the Head of Product and Chief Data Scientist, I developed the new checklist concept based on insights gained from our extensive user research. I built low-fidelity mockups in Balsamiq and a set of comprehensive interactive Axure prototypes to refine the concept and gather new insights into the evolving product. Continuous research and feedback helped guide our development process, but it also provided opportunities to help manage expectations and build a better understanding of our progress with customers.

Many different ideas and approaches were tested and either abandoned or modified into other components. For example, I quickly realized that we needed to move away from a nested folder tree after learning about the deep folder structures used in the insurance industry.
The interface responds to the selection the user makes by displaying thumbnail previews and presenting only the appropriate controls.

Our research revealed that policy checks often involved multi-screen workstations, which inspired a new experience that synchronizes checklist selection, document scroll, and highlights differences (i.e. redlines) across browser tabs and computer monitors. Through iterative improvements, we achieved an MVP that accelerated policy checks by 70% (according to a customer's internal audit).

This policy check shows documents being compared in the same tab, but this functionality works across browser windows and monitors. Selecting a checklist item aligns the document viewers to make comparing the contents easier. The "show differences" (redline) feature makes it much easier to visualize the differences between the two documents.
I maintained specs for the UI components in an Axure design library that was built as a forerunner to a more formal system.

Adapting to COVID

While the visual design system was developed alongside the product, building out a formal component library was impacted by the arrival of COVID-19, which shifted product priorities to address emergent risks related to the pandemic. We adapted by designing and building a new interface for our supervised review tool, enabling the efficient management of a very large influx of documents. 

Selecting items from the list scrolls to the content and highlights it for review. Users also help train the algorithms by modifying selection results when the system misses values.

In close collaboration with the Lead Developer, I rapidly designed a new interface for our supervised review tool that was built within a single sprint. We were able to accomplish that thanks to my use of proven design patterns and the intentional creation of reusable components. Throughout the process, we continued to enhance the document review management console, refining our toolset until RiskGenius was acquired by Bold Penguin, integrating it into their suite of insurance tools.

The document processing grid provides statistics and many different views into the work being done by the supervised review team.

Conclusions

This experience reinforced the importance of working closely with users and maintaining an ongoing feedback loop, resulting in a product that met unspoken needs and exceeded expectations. It also highlighted the importance of flexibility throughout the development process, both in terms of approach and the design of the product itself.

This was an extremely valuable learning experience for me, thanks largely to the successful introduction of a new product direction and creating an innovative solution from scratch.

Check out my other examples or learn more about me.

Get in touch via email or connect with me on LinkedIn.