Ownly, Sole Product Designer.
Two years designing the self-serve path to buying a new home: discovering, configuring, verifying, and reserving without a sales centre. The most useful thing I learned came from getting one decision badly wrong.
The legacy path to buying a new home is manual, time-bound and opaque. Buyers work through a sales centre, wait on pricing that arrives days later, and repeat the same conversation with three different people. Builders never hear from anyone who showed up after the doors closed, and have no idea how many there were.
The opening was a self-serve journey that worked at any hour. The difficulty is that removing the salesperson also removes the person who handles doubt. Every reassurance a rep would have given in the room had to be designed into the interface instead, across three products: ReserveNow (hold a home online), Verified (prove you can actually buy it), and Build & Price (configure it and see the number move).
Verified asks buyers for identity and financial information, so I assumed the friction lived in those steps, and that's where nearly all my early effort went. Making each step feel low-stakes. Explaining why every field was needed. Cutting the number of screens.
What I hadn't questioned was when a buyer was meant to do it. Verification stood on its own, as something a buyer had to decide to go and start, on the reasoning that anyone serious would seek it out when they were ready.
Of the buyers who started a reservation, 75% never found it. Not abandoned partway. Never found.
That figure measures discoverability, not drop-off. The people it counts didn't reach the steps and give up on them; they never learned the steps were there. Most found out verification existed only when a builder raised it directly, by which point it read as an obstacle someone had sprung on them rather than a step they'd chosen. The screens I'd spent weeks polishing were never the problem. Almost nobody saw them.
So I stopped treating verification as something to be found and made it the thing that follows reserving. A buyer picks a home, reserves it, and verification arrives as the step that completes the commitment they have just made. Same steps, same fields, asked at the one moment the buyer has already decided.
Once buyers reached it, Verified did what it was built for and shifted the first builder conversation from qualification to confirmation. Builders stopped spending early calls working out whether someone could actually buy, and buyers arrived knowing they could.
The three products shipped with builders including Shane Homes, Calbridge Homes, Logel Homes and Westcreek, each running a different sales process on the same platform. That variation was the real design constraint: a flow rigid enough to be trustworthy on the largest purchase of someone's life, and loose enough that four sales teams could recognise their own process inside it.
Measured across buyers who started a reservation, verification drop-off fell from 75% to 45%. It didn't go to zero, and the 45% that remains is a different problem from the one I fixed: those buyers find verification and choose not to finish it, which is a question about what the process asks for rather than whether anyone knows it exists.
Build & Price was the piece that changed buyer behaviour most visibly. Configuring a home used to mean a spreadsheet quote that came back days later; now the number moves as you choose. It reduced quoting delay to nothing, and buyers who configured a home converted at a noticeably higher rate than those who didn't.
Try it: change an option and the price updates live
Discoverability failures look exactly like usability failures in a dashboard, and they are not the same problem. If I'd only read the completion rate for Verified, I'd have kept optimising steps nobody was reaching. The lesson wasn't "test earlier"; I did test. It was to check whether people can find the thing before spending three weeks making the thing pleasant.