DriveAssist

DriveAssist had to hold a four-figure inventory of cars and still feel like a curated storefront, not a classifieds board. I collapsed the catalog into a Make → Model → Generation → Body structure, so discovery starts from the car people picture in their head — not from a wall of filters — and every model card stays ready to sell.

I connected the services around it — buying, selling, financing, equipment, service — into one hyper-linked system: wherever a customer enters, every next step is one link away, with no dead ends between services.

A car purchase is a web of interdependencies: with a trade-in, the final price is unknown until the old car passes appraisal — and until then there is no credit, insurance, or delivery to configure. My answer was to split the flow in two: ordering, where the customer defines scope in a few clicks, and processing, where appraisal, financing, and paperwork run over days with visible statuses.

I designed the order as a set of components — appraisal, credit, insurance, delivery, payment — each with its own status and structure, so the customer stays informed of every change and can amend, upsell, or pay online at any point.

I designed within a system of variable breakpoints: components respond independently instead of snapping the whole page, which keeps layouts alive on any resolution. The same system carried the quieter craft I cared about most — a chameleon header that adapts its color to any hero imagery, a curated car-color library that keeps catalog cards neat, and an email and SMS guide so even a payment notification speaks in the service's voice.

The platform ran as a product with a feedback loop, not a website that ships once. An evaluation plan defined what, how, and when to measure — do people buy, book, and appraise cars; where the main flows drop off; whether the experience feels right. A measurement plan wired analytics into every screen, insights fed the roadmap in a continuous research loop — and my design decisions answered to those numbers.