Estimate tool simplification (+300% completions)
Work at Marble.com (All Granite & Marble Corp). Numbers are approximate. Internal details omitted out of respect for the company.
The online estimate tool went through three generations over roughly a decade. I was involved in all three.
How it started
The first version was Flash — I owned the UX, coordinated the build, reviewed the API docs, and made sure the ERP pricing pulled correctly. When Flash started losing browser support around 2013–2015, I pushed for a migration before it became a crisis.
Version two was Angular. Same role on my side. When Angular’s own support cycle ended years later, another migration became unavoidable.
For version three the developer chose Pixi.js — that wasn’t my call. But the project stalled, and I stepped in to lead it through to completion.
What I actually changed
Every version had carried over too much complexity from the one before. Too many shape options, too many steps, too much that needed explanation. I watched real customers come into the office and struggle with it. Hotjar confirmed what I was seeing.
The core fix was simple: 99% of marble jobs are a rectangle or an L-shape. I cut everything else. Two shapes, cleaner steps, nothing that didn’t need to be there.
I checked every decision with the sales team — they knew exactly where customers gave up. And I kept watching real people use it until the drop-offs stopped.
One deliberate choice: email and phone at the end stayed optional. No gate.
What happened
Completions jumped +300%, quickly. People started coming into the office with the emailed estimate already in hand — that hadn’t happened before. More of them left their contact details voluntarily, even though we never required it.
What I took from it
Watching someone actually use the thing is worth more than any stakeholder meeting. And “fewer options” is almost always the right answer when the task is narrow — the edge cases aren’t worth the friction they create for everyone else.