A Major Automaker Disrupts Its Own Vehicle Search
Role: UX Lead for Inventory Experience design.
Artifacts: Wireframes of car-finder UI, Wireframes of dealer contact form UI, user personas for UX and development, asset usage guide, research synthesis doc.
The Problem
Synopsis:
As part of a major brand overhaul, this well-known automaker wanted to disrupt its own online car shopping experience.
During design discovery, UX identified an unmet need: Giving customers a way to configure AND source their dream car online. As a result, we built an Inventory Tracker to integrate with the retail experience. What follows is a story of design-thinking strategy, and what to do when customers don’t quite know what they want.
We charged with the redesign of all inventory search across the principal site, the Fleet Sales site, and the Certified Pre-Owned (CPO) site.
CPO inventory search prior to redesign.
(Click to enlarge)
Within the then-current version of the main site and CPO site, inventory search was acknowledged as being completely insufficient by client stakeholders, dealer representatives and end-users. Feedback showed that shoppers were dissatisfied with the options given to them, and analytics showed high user attrition between the Car Builder UI and executing and inventory search.
The Work
Research Work
Web Analytics report and end-user (customer) feedback via surveys and face-to-face.
Competitor site analysis, and industry data supplied by client.
Workflow optimization studies with real end-users.
Rapid Lean UX testing sprints.
Design Work
Requirements framing based on input from end-users.
Co-Design sessions with stakeholders.
Lightweight wireframes, paired with functional prototypes.
Asset and pattern library creation.
Our findings showed that for a large proportion of the shoppers, the primary issue was the “needle in a haystack” phenomenon: There was no way for the system to surface a near-match for the shopper if an exact match for the car they built on the website was not in any nearby dealer’s inventory. By the same token, there was also no logic which allowed the system to search slightly outside of the geographic radius that a shopper had specified.
The solution we landed on was the concept of “fuzzy matches”, or including approximate results within returned queries. Then came the question of how much automation should be given to fuzzy matches. Because we wanted things to be as automated in the background as possible, we decided that the system could handle fuzzy matches two ways:
The system could presume flexibility within likely matches based on how much a feature’s importance was weighted. For instance, a shopper was more likely to be flexible on the interior color of the car than the exterior.
In cases where flexibility could be adjusted for things like distance, we included adjustable UI’s so that the shopper could let the system know how far they were willing to travel to get the perfect match. The system could then return them fuzzy matches which were slightly outside of their travel radius if they were closer to the shopper’s match than within it.
We then co-designed the UI with stakeholders using Lean UX methodology. With their help, we were able to build out a functional prototype in a very short amount of time. In usability tests, the fuzzy matching system intuitively made sense to end-users, and the transition from car building to inventory search was described as being 'seamless'. From this work we were also able to steer the UI pattern libraries as well.
The Result
The question was whether this would translate to less attrition and more direct contact between shoppers and dealers for test-drives. After 90 days, the data did indeed show that fewer users were abandoning the system, and that a larger percentage were making it all the way through to the test-drive message receipt confirmation page. The fuzzy matches were a success! Additionally, dealers in California and on the East Coast were reporting better quality direct leads through the new system.
The "fuzzy matching" system, post redesign.
(Click to enlarge)