Redesigned the buyer flows for bidding, margin estimation and checkout, fixed in Figma the problems that research and testing found, and specified them in detail for engineering.
Overview & Problem
Dealers across Europe buy used cars on this trade platform every day, but it was hard to use. Navigation was outdated, the buying experience felt different in each market, and the product had no research or design standards behind it. Everyday tasks, such as buying several cars in one order, were slow and confusing.
I worked on the problem from research to final design. I analysed 545 survey responses from dealers in 7 European markets, interviewed Polish dealers 1:1 in Polish, and reviewed the live platform, finding 37 usability issues. Our Team Leader built a working prototype with Claude, and I audited it (29 findings). Our UX Researcher then tested it with UK dealers, who completed 91.6% of tasks. I translated the findings from both the research and my audit into refined designs in Figma, and every decision was documented in Figma specs and Confluence, ready for engineering.
Research Across 7 European Markets
Before designing anything, I needed to understand how dealers buy across Europe. I started with the dealer survey and checked its data quality before drawing any conclusions.
Multi-Market Data Synthesis
Before trusting the survey, I checked its quality. 545 dealers answered from 7 of 8 target markets, but over 40% came from Germany, and Portugal and Poland had under 20 answers each. I treated both as early signals, not facts, and followed up with 1:1 interviews in Poland. I also stated what the survey couldn’t measure, so the results weren’t over-read.
1:1 Interviews in Polish
Poland’s survey sample was too small to rely on, so I interviewed Polish franchise dealers 1:1 in their own language. A guide prepared with marketing left room for open follow-ups, which surfaced 15 topics, from imports to fair listing ranking.
Turning Findings into Priorities
I linked each finding to a platform change and a priority: fix the basics first (complete car data, fair ranking, a mobile app version), then add growth features. The Product Owner called it the clearest research documentation they had seen.
Personas, Journeys & Usability Audit
With survey data and interview recordings in hand, I built personas, mapped the dealer journey and audited the live platform against Nielsen’s 10 usability heuristics.
Personas Built from Survey Data
I grouped survey answers into dealer segments and turned each into a persona. Shown here is Erica, a Dealer Principal. Most dealers like her buy 1 - 10 cars a month (63%), half work from rural sites, and their top priorities are reliable logistics, better support and transparent costs. Those priorities carry through the later design work, such as the required transport step at checkout.
Dealer Principal Journey Map
I mapped the Dealer Principal’s 5 steps on a vehicle page - what she thinks, does and struggles with - based on audits of 3 live platforms (UK, Australia, EU). Each pain point has a design opportunity; post-sale support is flagged for further research.
Usability Audit of the Live Platform
I reviewed the live platform against Nielsen’s 10 heuristics and logged 37 issues. Most fell into 2 groups: unclear system status (8 issues, 21.6%) and cluttered screens (7, 18.9%). This matched what dealers told me in interviews - the platform felt “poor, unintuitive, overloaded”.
Making the Case to Leadership
I presented the research and audit to leadership one problem area at a time, rated by severity. For example, unclear system status covered 8 problems: 1 catastrophic and 7 major. Turning user frustration into clear evidence helped secure buy-in for a full redesign.
Benchmarking, Flows & Specifications
Before sketching layouts, I studied 6 leading automotive platforms to learn which patterns work on a vehicle page and which to avoid.
Competitive Benchmarking
I compared 6 anonymised platforms: wholesale remarketing, a manufacturer used-car site, retail classifieds and 3 types of auction. Five proven patterns made it into our vehicle page: clear navigation, sharing and export, grouped data, prominent buy buttons and in-context help.
Purchase Flow for Multi-Car Orders
I mapped the full purchase flow across 5 roles: dealer, platform, payments, support and logistics. Dealers can build a basket of several cars and undo removals. Transport is a required step that fixes the invoice total, and each car can be paid differently - card or dealer finance. Checkout stays locked until every pound is assigned.
From Wireframes to Specifications
I reviewed low-fidelity layouts with product managers and engineers, then turned them into detailed specs. The vehicle page has 7 levels of information (L0 to L6). Level 4, the price panel, shows net price, 20% VAT, damages and VAT status separately, each with a short rationale for sign-off.
Vehicle Status Badges
I defined 4 badge types in priority order: damage and risk, price and value, sales type, and demand. Cards show up to 3 badges and vehicle pages up to 5, so screens stay easy to scan. A badge seen on a card always appears on the vehicle page too.
Designing for Edge Cases
I also designed for situations that break the default layout. For example, if a dealer selects a car make hidden under “Show more”, it moves into the short list, so an active filter is never out of sight (123 → 25 results).
Accessibility & Colour Contrast
I checked status labels, tables and buttons against WCAG 2.2 AA before handoff. The brand teal failed as text (1.92:1), so I set a rule: thin elements of 3 px or less use the darkest teal (4.59:1). All 6 status-label colour pairs pass AAA.
Auditing the Coded Prototype
Our Team Leader built the prototype in code, which let us test interactions early. Before it went in front of dealers, I audited it against our design tokens, usability heuristics and WCAG.
29-Point Prototype Audit
I reviewed the coded prototype and logged 29 issues: 18 high severity (must be fixed before testing or release), 1 major and 10 medium. They ranged from cart logic errors to a button that failed colour contrast.
Making Every Action Reversible
Four high-severity issues involved cart and checkout logic. I redesigned them in Figma so a dealer can switch an offer back to Buy Now in one click, can’t pay until every car has a payment method, and can’t add a reserved car to the basket - they can ask to be notified instead.
Testing & Final Design
Our UX Researcher tested the coded prototype with UK dealers on real buying tasks. I then turned the research and audit findings into the final high-fidelity screens in Figma.
Testing with UK Dealers
Six UK dealers completed 91.6% of tasks on the prototype, compared with about 48% on the old platform. Ease scores out of 7: multi-car basket 6.1, profit estimator 5.8, payment setup 5.4. Testing also exposed a split-payment logic gap, which I resolved in the Figma designs before release.
Final Design: Basket & Checkout
If a car in the basket sells, it is flagged and removed from the total. Adding a third car unlocks a 5% bundle price. At checkout, each car can use its own payment method - card or dealer finance - and the order can’t be placed until the full amount is covered.
Final Design: Clear Pricing & Profit
The vehicle page shows net price, 20% VAT and VAT status next to the bidding panel. After purchase, a profit estimator works out resale profit under the UK margin scheme, where VAT is due only on the margin, and shows the break-even price.
Reflections - Product Work Beyond the Screens
This project showed me that good product design goes well beyond screens. The most valuable work happened between the steps: checking whether data could be trusted, auditing the prototype, and turning research and audit findings into design decisions before they reached engineering.
01
Documentation That Explains “Why”
Every key flow, edge case and state was recorded with its evidence - heuristic, benchmark or user need - so the team could see why each decision was made.
02
Working Closely with Engineering
Regular syncs with frontend developers resolved edge cases and kept the design system consistent, so handoff stayed clear and the build matched the design.
03
Turning Findings into Refined Designs
My prototype audit and the UX Researcher’s sessions with 6 UK dealers (91.6% task completion) pointed to the same fixes, such as a split-payment logic gap, which I resolved in Figma before release, not after.