From a stalled web build to 74 screens in three days.

Status: Client approved the design and flow. In build now, due in about a week. The review's main ask was more fields, not changes to the structure.

<aside> 🔗

See it running

• Interactive prototype: 74 screens at real phone size
• Design system & style guide </aside>

<aside> 🔨

In short

• The old web tool was built for a homeowner getting one estimate. My user is a roofer doing this five times a day. I designed a different product, not a smaller one.
• I made a job the centre of it. One address that collects a measurement, quotes, photos, notes and deliveries. The old tool had no idea of a job at all.
• I ran the developer's real calculator against my screens instead of trusting my own numbers. Most of my quantities were wrong, and three inputs the calculator needs were not being asked for anywhere.
• I removed five features I had already built, including one that was quietly rude to the user it was meant to help. </aside>

01. The Problem :

A buggy, half-built web tool and a five-line brief.

RoofBros is an Australian roofing supplier. Their developer had started a web tool where you type an address, it pulls the roof measurements from Nearmap satellite imagery, runs a materials calculator, and gives you a quote. It was buggy and unfinished, and some screens went nowhere.

So the mobile app was not a copy of the web app. It became the actual product.

All I was given was this:

"A roofer-facing mobile app (iOS + Android) that does the same on a phone, plus lets a logged-in roofer save and manage their jobs."

No flows. No wireframes. No screen list. I had to work out what the app was.

I went through the old web tool anyway, but not to copy it. I used it to collect the real data: tile profiles, colour names, materials, delivery fees. I left the structure behind, because it was built for the wrong person.

The web tool was made for a homeowner getting one estimate. My user is a roofer who does this three to five times a day, on a roof, in the sun, often wearing gloves, filling it in for a customer. Those are two different products.

02. What I Did With That :

One job, not one quote.

I made a job the centre of the app. Not a quote, a job, which is one address that collects a measurement, several quotes, photos, notes and deliveries over time. The old tool had no idea of a job at all. That one decision created most of the app.

here: the jobs hub (home), the pipeline filters a roofer opens every morning.

It also showed me what RoofBros actually sells. Anyone can copy a calculator. Not everyone can deliver recycled tiles. So the delivery request got the most attention: two ways to get materials (deliver to site or pick up at yard), because roofers do both and they need different fields. A delivery needs a drop-zone note and a photo. A pickup needs to know if you're bringing a ute or a truck.