A Product-First Approach to Roofing Workflow Management

A growing business where no system owned the whole job

As the sole product designer on this project, I led research, information architecture, and UI design; developers built to spec. “We” below refers to the wider project team on the decisions we made together.

Priority Roofing is a residential and insurance roofing contractor operating in the United States. Their business runs the full trade cycle, canvassing neighborhoods for storm and replacement work, estimating and closing homeowners, ordering materials, dispatching crews, passing inspections, and settling insurance claims and rep commissions. As volume grew, that operation outpaced the tools running it. A single job's truth was scattered across disconnected software, each good at one slice of the work, none aware of the others, plus whatever lived in someone's head. No one place could tell you where a job actually stood.

No system owned the whole job

The tools weren't the problem, the seams between them were. Every handoff meant re-keying the same job into another system, and every gap between tools was patched by hand. Four fault lines showed up again and again.

Following a job, not an org chart

Before designing a single screen, I traced one job end-to-end and sat with every role that touched it. The goal was to map how work actually moved from role to role and where it fell into the cracks between systems.

Decide what to own, what to keep, what to kill

The biggest early decision wasn't a screen. It was scope. Ripping out the whole stack at once would have been a political and financial fight we'd lose. So we drew a clear line around what the CRM would own.

One continuous path: knock to commission

The modules aren't a menu, they're a route. A lead enters in the field and leaves as a paid job, and every handoff between roles happens inside the same system. Four phases, each owned by the people who do the work.

Seventeen modules, three groups, one job at the center

The architecture follows how the business thinks, not how software is usually organized. The navigation splits into three groups, the daily work, the money, and the people, each pointing back to the job.

Material management was the heaviest lift. Inventory, purchase orders, and reconciliation all ran by hand across tabs. Alongside fees, sales reps, PMs, and territory, each spreadsheet was replaced by a structured module tied back to the job.

A Prospect is the module for the whole Acquire → Qualify stretch, from a captured lead through status changes to Job Won. It's the busiest screen for Sales Reps and the gate Back Office checks before a submittal becomes a real job.

Every job runs on an eight-stage pipeline pinned to the top of its record. It replaced the Excel status column no one trusted: the current stage is unmistakable, each stage has a clear owner, and "Next Step" is a first-class field instead of something you deduce.

QuickBooks stays the book of record, but the operational money that used to live in Excel (commissions, draws, and material spend) now sits inside the CRM, tied to the job that generated it.

I built the design system to create a shared foundation for the product and reduce the time spent solving the same design problems across screens. Reusable components, patterns, and tokens helped me streamline repetitive tasks while keeping the experience visually consistent and making it easier to maintain as the product evolved.

It also improved how I worked with developers and stakeholders. With clearly defined components, variants, and usage guidelines, I could communicate design decisions more clearly and hand off work with less ambiguity. This created better alignment between design and development, reduced inconsistencies, and made future changes much more efficient to implement.

A system this central to daily work couldn't be judged on adoption metrics alone. Ahead of the full rollout, we ran a structured pilot with the three roles that live in the CRM every day, then used what came back to decide what shipped as-is, what got reworked, and what got cut before it reached everyone else.

Before and after a two-week pilot, we asked every Sales Rep, Project Manager, and Back Office user to self-rate four things, 1-5. The gaps told us where the design was working and where it wasn't.

Three tools became one, and the work became visible

The spreadsheet dependency that patched every gap between tools disappeared. For the first time, a job's whole life (lead, production, and money) was legible to everyone who touched it.