Case Study
Optimizing self-serve claims tools to increase call deflection.
Role:
Lead Designer
Timeline:
2 months
Platform:
Web (Desktop & Mobile)
Context
Insurance clients can file a claim, also known in the industry as a "First Notice of Loss" (FNOL), through one of three channels: over the phone, via the mobile app, or online using web tools. Encouraging users to file claims digitally is ideal for the company, as it significantly reduces operational costs.
Problem
Fewer than 40% of users who attempted to file a claim through the online portal successfully completed it.
Goal
Increase the completion rate of online First Notice of Loss (FNOL) claim tools.
Result
Successfully increased FNOL completion rates from 39% to 54%.
Research
Initial research was conducted through user interviews, usage recordings, heatmaps, and data analysis. While legacy systems presented some constraints on filtering data, we were able to gather meaningful insights to inform our decisions.
Relevant User Feedback
User feedback, gathered through surveys and discussions with call center agents, highlighted potential interface ambiguities and form timeout problems, while also suggesting that exploratory behaviors might be contributing to the bounce rate.
“It’s not letting me submit my claim.”
“I didn’t want to screw up and have my claim denied”
“I needed to go get my files, and when I got back the claim had been lost.”
“The button is not working”
“I wasn’t sure how to answer the question”
“I was just curious what the form looked like”
Drop off in the auto claims form.
Analyzing the completion funnel, we identified four key steps where user drop-off was significantly higher:
Step 1: Vehicle
Step 2: Accident Type
Step 7: Location
Final Step: Contact Information / Submit
Digging into step 1: Vehicle
To understand how much of the 18% bounce rate should be attributed to usability issues versus user exploration, I requested the data team filter out users who did not open a claim—digitally or over the phone—within two months of visiting the page.
Unfortunately, this analysis was not feasible due to siloed and legacy databases. However, user session recordings uncovered clear usability and accessibility issues that were contributing to the problem.
Unfamiliar interaction patterns and insufficient guidance hindered usability.
The FNOL form introduced an unfamiliar accordion pattern, with upcoming steps visible but disabled until the previous step was completed.
This caused confusion in the Vehicle step, where over 90% of users had only a single vehicle on their policy. Although the vehicle was displayed, it wasn’t preselected, and unclear wording left users unsure they needed to click it to proceed.
Lacking guidance from poor error handling, many users mistakenly tried to interact with the disabled second step, perceiving the platform as unresponsive and resulting in frustration and drop-offs.
Digging into step 2: Accident Type
Initially, we believed qualitative data would explain the high drop-off rate. Interviews revealed that ambiguity in the options left users feeling they couldn’t accurately tell their story, leading some to abandon self-serve channels.
However, the most significant discovery came from the quantitative data -or rather, from uncovering inaccuracies in reporting.
Initial reporting did not account for glass claims deflection.
The "Glass Damage" option directed users into a separate claim flow, accounting for 6% of interactions and lowering the actual drop-off rate from 8% to 2%.
With this insight, we deprioritized optimizations for this step and roadmapped future improvements based on qualitative findings.
6% of users redirected.
Adjusted drop rate: 2%
Digging into step 7: Location
Analysis of user recordings revealed a substantial drop-off rate at the "Address" field in the location step. Quantitative data from the mobile app’s equivalent step showed only one-third of the drop-off rate compared to the web platform, emphasizing a potential issue specific to the web flow.
Users felt unable to provide accurate information in a required field.
Follow-up interviews revealed a key issue: many accidents occurred on unmarked stretches of roads or highways, making it challenging to provide a specific address. The problem was compounded by the web platform's mandatory "Address" field and overly prescriptive label, which pressured users to enter nonexistent details.
Data from the mobile app's equivalent step supported these findings. Its more flexible design, including a less rigid field label and the option to skip the step, resulted in a drop-off rate only one-third of that seen on the web tool.
14% Drop
Mobile Web
5% Drop
Mobile App
Digging into Contact Info & Submit
Of the users who reached this step, 16% (7% of total funnel users) dropped off instead of clicking “submit,” despite their contact information being pre-populated. User feedback suggested the submit button appeared unresponsive, and session recordings revealed users struggling to understand why it was disabled.
Accessibility issues amplified by poor error handling and weak visual hierarchy
The use of disabled buttons instead of contextual error states created significant accessibility challenges for users who missed earlier questions and encountered a disabled submit button at the end of the form. The low-contrast design failed to meet WCAG guidelines, causing many users to miss the button entirely. For those who noticed it, the unresponsive button and lack of guidance left them confused and frustrated, driving many to abandon the form.
Solutions
I identified key opportunities for improvement, projected their potential impact based on previous research, collaborated with engineering teams to validate assumptions, and worked with stakeholders to secure buy-in and inform roadmapping decisions.
Update location field label & requirements.
Address accessibility issues & error handling.
Auto-select vehicle when applicable.
Transition from a web-form to a wizard.
Update location field label & requirements.
I recommended four small optimizations, projecting a 10% increase in step completion and a 6% improvement in overall FNOL completion (based on data from the mobile application’s location step).
Update the field label to “Location” to convey greater flexibility to users.
Revise the placeholder text to better align with the use cases identified in user research.
Increase the field height and add a character counter to encourage descriptive input.
Enable the primary CTA, improving accessibility and allowing users to skip the field if needed.
Before
After
Address accessability issues & error handling.
I recommended updating the error handling pattern from disabling the submit button to a more accessible approach, enabling the button while providing clear guidance on fields requiring correction. Additionally, I proposed improving the visual hierarchy of the submit button to enhance visibility and introducing a confirmation modal to prevent users from accidentally exiting the form before submitting.
Errors Before
Errors After
Auto-select vehicle when applicable.
To streamline the process for over 90% of users, I suggested adding logic to the vehicle selection step that would automatically preselect the option for users with a single insured asset.
Transition from a web form to a wizard.
Recognizing the limitations of our FNOL web form (and web forms in general), I began championing a transition to a wizard-based experience. By breaking the process into guided, manageable steps, this approach aimed to reduce cognitive load, improve error handling, and enhance accessibility.
9/10 users preferred this format.
I built a medium-fidelity prototype and conducted usability testing with 10 users. Nine participants found the single-question format more approachable compared to the existing web form, validating its potential to improve the user experience.
1 user did not.
The tenth user expressed concerns about the ambiguity of the experience's length, prompting us to add a research initiative to the backlog that focused on the progress bar.
Bandwidth limitations delayed this initiative.
While the value of transitioning from a web form to a wizard was acknowledged, bandwidth constraints required us to prioritize quick wins and roadmap the project for a later date.
However, I leveraged an upcoming initiative to optimize the FNOL experience for property claims, successfully collaborating with the product team to expand the scope and include the transition to a wizard for this tool.
Results
Overall increase: 39% to 54%
(and that’s just from the quick-wins)
Auto FNOL had its quick-wins go live.
Solutions 1 through 3 were implemented, increasing the FNOL form completion rate for auto claims by 15%.
Wizard redesign slated for Q4.
Before the transition to a wizard was initiated for auto claims, I left Intact Financial Corporation to lead the design process at Wethos.