Insurtech: UX strategy

Insurtech Case Study: Building Trust 

sayscreens.png
 
 

The Challenge

Say Insurance is a young insurtech company providing online auto insurance. Like any new player with a field of established competitors, developing and maintaining trust is an essential part of gaining market share. My primary roles on this project were UX strategist and researcher.

The Process

I researched Say's website using a heuristic evaluation, competitive analysis and streamlined cognitive walkthrough to find ways that the company could improve usability, a known credibility-booster. I then tested a key online feature, the insurance score, in remote unmoderated interviews and a task-based survey to get feedback from users.

I combined analysis from additional academic research on building online trust, insights from the interviews, and quantitative data from task-based survey into a video presentation with key takeaways to inform future designs. 

This work was performed as student in the MHCID program at UCI.

Discovery

Standing Out: Heuristic Evaluation

One of Say’s differentiators is a feature called Insurance Score. This feature allows users to learn about a relatively unknown but important factor that went into their rate. I led a heuristic analysis, working with three other MHCID students, based on Nielsen’s usability principles.

While the site met many usability criteria, my report found room for improvement on user control, consistency in the menu, and implementation of error prevention. Only one feature was ranked severe: users were required to give their address before learning if they're out of area and unable to receive their insurance score. 

Building Trust: Competitive Analysis

In insurance, you don’t actually see the product being sold. Instead, the customer is purchasing security and peace-of-mind. I performed a competitive analysis on three direct and indirect competitors, plus three innovative companies from other verticals. My goal was to learn ways Say could stand out as a trustworthy new player. Part of the analysis included a light heuristic evaluation of the competitors. I used Peter Morville’s “honeycomb” of usability heuristics this time because of his consideration for credibility as a key heuristic. I also studied published academic research on online trust, credibility, and security.

Proof of insurance

To learn how user-friendly the site might be for current customers, I performed a streamlined cognitive walkthrough on the process of obtaining proof insurance. I chose this scenario because all drivers are legally required to show proof of insurance when registering their vehicle, among other common situations, so it affects all of our users.

 Excerpt from the streamlined cognitive walkthrough report

Excerpt from the streamlined cognitive walkthrough report

Delivery

I paired qualitative research with quantitative data from users to get additional insight into what users were doing and why they were doing it. I tested the insurance score flow through three unmoderated user interviews with UserTesting and a task-based survey with over 70 screened participants with Loop 11. I created this final video presentation that combined all of the research findings into strategic and tactical takeaways.

Challenges overcome

Can I have the last four of your social security number?

Testing a feature that requires personal information was the most difficult part of this project. I couldn’t (nor did I want to) have testers give their personally-identifying information, but I wanted to hear and see their real-life reactions to being asked to provide highly sensitive data like the last four of their social security number.

My workaround

 Trial and error while finding the right tool for the task

Trial and error while finding the right tool for the task

First, I attempted to layer Loop11, the task-based survey tool, on top of a clickable prototype within InVision. The prototype looked just like the site but would allow testers to progress through the flow without giving any information, yet still be interrupted with questions like “Would you give your address on this page?” before moving on. However, layering InVision’s on top of Loop 11 created such a loading lag that testers were abandoning.

I knew I wanted the clickable prototype, but InVision wouldn’t cut it. So I started over, this time in JustInMind. I recreated the live site by importing screen-grabs of each site page into JustInMind, linked it all together using hotspots, and created a new clickable prototype. I exported the HTML of the project and deployed to Github. Layering Github with Loop11 resolved the laggy load. I also used that same coded prototype to perform the unmoderated user interviews with UserTesting.