heycar · B2C · 2022–2023Cutting heycar's search from five fields to three
High-intent shoppers were abandoning heycar's homepage search before they ever saw a listing. I cut it from five fields to three and shipped it across four European markets, as part of a redesign that lifted regional customer engagement by 18%. Three years later, it's still the live search.
The problem
The homepage asked too much, too early. Five fields stood between a visitor and their first look at a car, and the funnel data showed high-intent shoppers arriving and leaving without ever starting a search. They weren't bouncing off the inventory. They never got that far.
What the research said
The bet I made early
Inside the company, more filters up front meant more value, and every field on that homepage was somebody's priority. I bet the other way: the homepage's job is to make starting cheap, and the depth belongs on the results page, where you can actually see what your choices do to the results.
To keep that argument from turning into a matter of taste, I got the team behind one metric: search-initiation rate. Not engagement, not time on page. Do more people start a search, yes or no. Having a single testable claim is what carried the design through every stakeholder review that followed.
The design
The new search asks for three things a shopper already knows when they show up. Everything else moved to the results page as progressive filtering, where narrowing down feels like control instead of homework.
Together with two other designers, I defined the tokens, interaction states, and accessibility behavior for the new components, built so all four markets could share one implementation. The per-country variants were retired.
Before and after: the user's journey
A high-intent shopper's path to their next car, before and after this work.
Land on the homepage
Arrive with clear intent, greeted by five inputs: make, model, city, radius, and a price toggle.
"I just want to see the cars." The tool meant to help is the first obstacle.
Three fields: make, model, zip code. One unambiguous CTA.
"This will take ten seconds."
Start a search
Five decisions before seeing a single car. Five chances to second-guess and click away.
Hesitation. Many high-intent visitors left without doing anything, in every market.
Three inputs they already know the answers to.
Momentum. Starting a search costs almost nothing.
See results
Many never got here. Those who did had set filters blind, so results reflected guesses.
Frustration or absence. The funnel dropped people before the product could show its value.
Results appear fast, and now there's something concrete to react to.
Progress. Real cars, real prices, something to refine.
Refine
Refinement happened up front, without context. Users who set radius or price on the homepage changed them again on the results page anyway.
Redundant. Work done twice, once blind and once informed.
Radius, price model, and advanced filters live on the results page, where users have context to use them.
Natural. Narrowing down, not filling out a form.
Return later
Four markets, fragmented per-country implementations.
Unfamiliar. The product felt different depending on where you were.
One shared search pattern across all four markets, built from design system components.
Consistent. Still live three years later.
What actually changed
Nothing was removed, everything was sequenced. The questions users couldn't answer on the homepage moved to the results page, where they could. The entry point stopped taxing intent and started converting it.
Outcomes
More work
Have a project in mind? I'd love to hear and connect with you. Let's get in touch and discuss how we can bring your ideas to life. Email me at nr9473@gmail.com