Brainsome
Product design · 2023–2025
Requote
A clearer insurance journey, from early research to a live product.
- Role
- Product Designer
- Team
- PM, Data, QA, 4 Engineers
- Scope
- Research, UX/UI, experiments
- Stage
- Research through production
What began as Car Insurance Research became Requote: a live, mobile-first insurance funnel. I worked from early form concepts through product iteration, improving lead value while keeping a long journey understandable and trustworthy.
01 / Retention
Continue where you left off.
+100% returning-user offerwall clicks.
+8% overall EPC.
Traffic analysis and feedback showed that roughly one in five visits came from returning users. They still had to complete the entire form again — even when very little had changed.
I designed a return path to review previous information, edit individual answers, restart or restore earlier offers. Existing cross-vertical data could prefill part of the profile, but users still needed a clear opportunity to confirm its accuracy.
The flow separates recognising a previous visitor from recovering their quotes. If the browser no longer has the earlier session, the person can look up their details instead of losing access to the return path.
Retention / Entry & recovery
Different starting points, the same way back.
02 / QuickMatch
Quotes without the interruption.
+12% overall EPC.
+120% offerwall clicks.
Usability testing showed that the popup after vehicle entry covered most of the screen and interrupted the full journey. It generated clicks, but could pull people away before they reached the more valuable final offerwall.
I moved QuickMatch offers to a dedicated page. The header first shows progress towards an initial quote, then becomes “My Quotes” with the number of available offers. A lightweight notification announces new matches without blocking the current question.
Overall clicks fell 4% while earnings per click (EPC) improved 12%. Fewer low-value QuickMatch interactions were an acceptable trade-off when more people continued to the final offers.
QuickMatch / Before & after
Keep progress in the same place.
03 / Processing & calls
Another way to get help.
Calls contributed over 5% of monthly revenue.
Analytics revealed revenue potential in phone-based leads. Working with Business Development, we added a dedicated partner number, but the partner required a complete profile and calls lasting over 30 seconds.
This added processing complexity before the offerwall. I made the intermediate states clearer and introduced the call option without removing the route to online quotes for people who preferred not to call.
Form-to-offerwall conversion remained nearly unchanged, with a drop below 1%, while calls grew to just over 5% of monthly project revenue.
04 / Offerwall
Make comparison easier.
+18% form-to-click conversion.
+4% EPC.
Seven or eight nearly identical partner cards made choosing difficult. Session recordings showed hesitation, slow scanning and rage clicks, while delayed offer loading added another source of friction.
I simplified the cards, brought the comparison information forward and made the main action easier to find. Together with the team, I also worked on sorting and loading behaviour, then highlighted the recommended offer.
The redesign improved conversion from form submission to an offer click by 18% and EPC by 4%. Further iterations on presentation, sorting and recommendation produced another 8% more clicks and 3% more EPC.
05 / Research foundation
Before the live product.
From form exploration to an editable, continuous journey.
Car Insurance Research: early concepts and wireframes for Requote.
I began with desk research into the U.S. insurance journey, competitor flows, proto-personas and job stories. The aim was to understand what the form really needed to ask — and where people needed more control, reassurance or a way to save their progress.
The main needs differed: first-time buyers needed guidance; households needed multiple cars and drivers; experienced customers wanted less typing and a faster route through familiar questions. This led to a modular flow with editable answers instead of a single, irreversible questionnaire.
I shared standalone concept screens with partners and potential users in interviews, before investing in an interactive prototype. Progress indication helped people orient themselves, and support for several vehicles was especially relevant.
Other ideas became less important after feedback. Inline explanations were appreciated but rarely requested spontaneously; a time estimate was useful rather than essential; and partners confirmed that ZIP entry was sufficient without automatic location detection.
Real-time policy pricing remained a hypothesis. People were interested but cautious about trusting it, and without direct insurer integrations we could not provide actual prices inside the form. That part of the concept was not implemented.
The three wireframe directions tested different ways to manage complexity: one question at a time with an answer summary; a continuous conversational history; and grouped questions with persistent section navigation on desktop. The conversational version was a lighter form, not an attempt to imitate a human agent.
The production work above developed the same themes further: preserve answers, make progress understandable and let people choose how to continue. Across a year of active development, the team tested and iterated on more than a hundred hypotheses; these four production chapters are selected examples.
