Research Phase:
Users abandon long forms 2–3× more often when steps are static and overwhelming.
Short, guided steps dramatically improve completion rates. Breaking tasks into short screens with progress cues reduces cognitive load and keeps attention high.
Sources: Baymard Institute (2023), NN/g Form Usability (2020), Google UX Playbook (2022)
Pain Points
I conducted three remote interviews with participants aged 22–32.
Younger users want personalization and have lower tolerance for dated UI and data sharing.
“I finally found a site with the info I wanted, but I left because it looked sketchy.”
“I'lI just want to see if I’m interested before handing over all my info.”
“I’ve used agents before because they actually personalize the quote for you.”
Analysis
Tools for an insurance site to behave like any other online shopping experience
Design Direction
Research revealed 4 tools for an insurance site to behave like any other online shopping experience
Clarity
Users couldn’t understand coverage without guidance. Confusing terminology and early data walls made exploring plans feel risky instead of safe.
Guidance
People expected the flow to feel like talking to a knowledgeable agent, not guessing through long forms and unclear progress indicators.
Trust
Credibility was judged in seconds. Outdated UI and unclear pricing made users question legitimacy and abandon before exploring options.
Efficiency
Users wanted to compare plans like online shopping. They wanted fast filters and side-by-side clarity, not digging or calling to understand differences.
Iterations + Reasoning
Driven by user testing and feedback
Final Designs
Outcome :
PolicyWise is now positioned to operate as a B2C platform.
PolicyWise can now meet the expectations of younger users seeking fast, intuitive filtering without needing to contact an agent, overshare personal details, or decipher jargon.
Key Takeways
This was my first major end-to-end design project, each iteration taught me something new.
Balance openness with structure in research.
The card sort surfaced a clear mental model (“I know / I don’t”), but the open format led to inconsistent groupings. Next time, I’d use a hybrid sort. This would keep space for insights to emerge while adding enough structure to produce usable patterns.
Prioritize by impact, not interest.
As a student, I tried to fix everything. In a real product environment, I’d sequence work by value: first clarity (reduce jargon), then success (filtering logic), then trust (emotional reassurance). This ensures effort is focused on the most critical usability issues.
Expand validation to support long-term confidence.
If revisiting today, I’d test AI-assisted recommendations, verify accessibility for screen readers and keyboard flows, and use tools like SUPR-Q to measure trust and loyalty (not just task success) since users signaled trust as a key barrier to adoption.
