Skip to main content
keystoneloom
Rebuild studySummit Ridge RoofingRoofing

A roofing website rebuilt around one goal: the booked estimate.

The baseline is a site we see constantly in audits: a heavy theme, a buried phone number, no service-area structure. The rebuild shows what a roofing site looks like when every decision serves the estimate request.

A rebuild study: a typical underperforming website, rebuilt to our standard under a fictional brand. The engineering is real — every number is measured on the build.

7.2s → 0.8s

Mobile load, baseline vs rebuild (lab)

Archetype

Storm-market roofing contractor on an aging template site

Build time

6-week rebuild

The starting point

The baseline models the roofing site we audit most often: a purchased theme carrying about 4MB of scripts and sliders, a 7.2-second load on mobile data, the phone number hidden in a footer, one generic "services" page for every kind of work, and no pages for the cities the company actually serves. It looks busy and performs like a brochure left in a drawer.

The rebuild

We rebuilt it as we would for a paying client. A tap-to-call header and an estimate request on the first screen. Separate pages for repair, replacement, and storm work, plus a service-area structure Google can read. Insurance and financing questions answered in plain English, and a project gallery that loads fast instead of impressing slowly.

Under the hood: hand-coded pages, structured data describing the business honestly, and a performance budget enforced in code — the mobile load dropped from 7.2 seconds to 0.8.

What it demonstrates

This is the exact playbook we apply to real roofing sites: conversion path first, local-search structure second, speed as the multiplier on both. The numbers below are measured on this build — and they're the same numbers we'd measure on yours, before and after.

Measured on the build

7.2s → 0.8s

Mobile load, baseline vs rebuild (lab)

100

Lighthouse performance, mobile

−88%

JavaScript shipped vs baseline

1 tap

From any screen to a phone call

Measured on the rebuilt concept in lab conditions: mid-range Android device, throttled 4G.

Before / after — measured on the build

[ Baseline ]

7.2s

Mobile load (lab)

  • 4MB of theme scripts and sliders
  • Phone number hidden in the footer
  • One generic services page
  • No service-area structure

[ Rebuild ]

0.8s

Mobile load (lab)

  • −88% JavaScript shipped
  • Tap-to-call from every screen
  • Repair / replacement / storm pages
  • City structure Google can read

100

Lighthouse performance, mobile (lab)

Get a straight answer about your website.

Send your URL. We'll tell you what's costing you customers — specific findings, plain English, no obligation.