← All SELECTED WORK
Rebuilding car finance for ClearScore's new marketplace
The 30-second version
Role
Product design, user research, prototyping, stakeholder alignment
Team
1 product designer, 1 product manager, 4 engineers, 1 EM
YEAR
2026
Platform
Web & mobile
PROBLEM SOLVED
ClearScore moved all its credit products onto a new platform in 2026. I designed the car finance product vertical, all four journeys for people buying a car or refinancing one.
+10.9%
more people clicked through after viewing their offers
+15.5%
revenue per person who viewed offers
+17.4%
for HP, more people went from logging in to clicking an offer
SCOPE OF WORK
ClearScore helps 16 million people in the UK check their credit score and compare credit. After credit cards and loans, car finance is one of its main revenue drivers, so any drop in performance would be felt quickly.
I designed ClearScore’s car finance experience end to end as it moved onto the new marketplace: the product picker users start from, the routes between them, and the offer pages they end on.
Car finance covers four sub-product verticals:
Hire Purchase (HP): buy a car and pay it off monthly until you own it.
Personal Contract Purchase (PCP): lower monthly payments, with an optional lump sum at the end to keep the car.
Switch your car loan: move an existing car loan to a better deal.
Finance your final PCP payment: spread the cost of that end-of-deal lump sum.
WHAT IS THE PROBLEM WE'RE SOLVING?
Migrate the ClearScore car finance product vertical into the new marketplace in the new design system, without changing how it works.
Success meant three things:
Nothing breaks: every question, rule and branch behaves as before.
Lenders get exactly the same data.
Revenue holds against the old journey.
That sounds simple. It wasn't. The four journeys send people between each other, each ends on a different set of offers, and years of rules lived only in the old screens. Missing one could quietly cost money or break what lenders receive.
72% of ClearScore potential car finance users were driven by affordability, not product choice.
VALIDATION RESEARCH
HOW I APPROACHED THE PRODUCT DESIGN
Before designing, the old journeys were documented completely. I wrote a primer on how each product works, audited every screen, question and rules across the four journeys and their offer pages, mapped every answer to the new platform, and raised open questions early.
I sat with engineering to refine the existing logic in implementation mapping sessions. Outdated requirements were flagged, as well as use cases that made easy car finance application between the different routes complex.
Journey map — Old car finance
Notion documentation of each journey requirement
The audit turned up problems nobody had written down:
Tangled routes. People who picked PCP without a registration number were moved into hire purchase with no clear signal.
A dead end in balloon journey. People with an older car could get stuck with no way back.
An unexplained phone number. Hire purchase asked for one without saying why. On investigation, Zuto, one of the lending partners, needed it to contact people about their eligibility. This had to be consented to by the users.
Inconsistent checks and errors. Each journey validated answers and showed errors differently.
Unclear card logic and navigation. Nobody could fully explain when cross-sell cards appeared. On the refinance journey, the back button went to the landing page instead of the picker.
DESIGN DECISIONS
Design the routes between journeys, not just the journeys
From the car finance picker, people take one of four routes: hire purchase, PCP, switching a car loan, or financing a final PCP payment. Each has its own questions and its own offers, and needed to stay clear from the first choice to the offers at the end.
Journey map showing for the four car finance routes
Validating input in the questions flow
Every journey is a series of questions, and the audit showed each one handled answers differently, even where the rules were the same. Some fields were checked as people typed, others only when they left the field or pressed next.
The rules couldn't change, but how they were shown could. Hire purchase still needed a phone number, for example, so I added a callout explaining why and asking for consent to be contacted by our car finance partners.
Personal Car Purchase questions flow
Update components in the design system
The design system already existed, so I matched every element in the old journeys to its closest component. Most mapped directly; a few, like PCP's live borrowing amount and the registration check, needed existing components combined to behave as before.
Component mapping: before and after
Offers page that fits each journey
Each journey ends on its own list of offers and offer detail pages. Offer screen is where people decide on which product to go for, and where the biggest gains came from.
Car finance offers page
READY FOR LAUNCH AND THEN MAYDAY…
We didn't switch everyone over at once. The new journeys went live to 1% of users in April 2026. At 5% we encountered our first issue.
A cadence gating bug made it impossible for users to edit their details before the 30 days which we prompt them to ensure their information was up to date. This caused a slight dip for a few days in the offers journey completion. A product and engineering decision was made to revert to showing the questions every time.
On resolving the gating problem, we rolled out to 30% and 50%, reaching everyone on 8 June. At each step we compared them with the old journeys on conversion and revenue before going wider.
IMPACT AND OUTCOME
Throughout the rollout, revenue per visitor showed no significant change against the old journeys, which was the bar for success.
TAKEAWAYS
Audit before you design. The most valuable design work happened before any screen was drawn: documenting every rule, route and edge case so nothing was lost in the move.
Rolling out in stages is how you catch problems early. The drop in people reaching their offers showed up partway through the rollout and was recovered before everyone was moved over.
Next