Performance and conversion are more closely linked on service websites than most teams realise. A visitor who arrived via a Google Ads click and encounters a three-second load time on a mobile device will often back out before the page finishes loading — and the cost of that click still lands on the advertising budget.
This case study documents a performance-focused rebuild that moved LCP from 3.0 seconds to 1.2 seconds on the key service landing pages, and the conversion improvement that followed.
The starting point
The business — a multi-service trades company operating across metro Melbourne — had a marketing site that ranked reasonably well organically but was underperforming on paid traffic. The conversion rate on Google Ads landing pages was lower than industry benchmarks, and the sales team's anecdotal feedback was that enquiry volume felt low relative to the ad spend.
A CRWeb Vitals audit identified the problem quickly. The service landing pages were scoring in the Poor range for LCP on mobile — the most common access device for the target audience. The primary contributor was the hero section: a full-bleed background image loaded at desktop resolution on every device, including mobile, combined with three third-party scripts that loaded synchronously before the page was visible.
Secondary issues included a DOM depth that was slowing rendering, render-blocking CSS from an outdated component library, and image assets throughout the page that were not sized or formatted for mobile.
What the rebuild changed
The rebuild was scoped deliberately. Rather than redesigning the site, the team focused on the performance bottlenecks identified in the audit.
Hero section overhaul
The full-bleed background image was replaced with a single foreground hero image, sized appropriately for the viewport, served in WebP format, and marked as a priority load. The preload hint was added in the <head> so the browser started fetching it before parsing the full page.
This single change accounted for most of the LCP improvement. A 2.4MB background image loading at desktop resolution on a mobile connection was replaced with a 180KB optimised foreground image that loaded in under 400ms.
Third-party script deferral
Three scripts were loading synchronously: a live chat widget, a booking plugin, and an analytics tag. All three were deferred until after the first user interaction.
The live chat widget was the most impactful deferral — it was adding approximately 800ms to time-to-interactive on mid-range Android devices. Deferring it until a user scrolled past the fold removed that cost from the initial load without affecting visitors who would have used the chat anyway (those users always scroll before initiating a chat).
DOM simplification
The page had accumulated markup from several iterations of design and plugin use. Sections were nested inside wrapper divs that served no visual or semantic purpose. Removing this extra nesting reduced DOM complexity, which improved rendering performance and also made the page easier to maintain going forward.
Image optimisation throughout
Every image on the key landing pages was audited. Images not in the initial viewport were marked with loading="lazy". Images that appeared above the fold were preloaded. All images were converted from PNG and JPEG to WebP where browser support allowed, and served at sizes appropriate to the layouts in which they appeared.
Results
After the rebuild, the pages were measured against the original baseline using Google PageSpeed Insights and field data from Chrome User Experience Report.
LCP improved from 3.0s to 1.2s on mobile across all five key landing pages. All pages moved from Poor to Good in the Core Web Vitals assessment.
On-page conversion rate increased by 22% on the primary service landing page over the 60 days following launch, compared to the equivalent 60-day period in the prior year.
Bounce rate dropped measurably on the pages receiving the most paid traffic. The most likely explanation is that fewer visitors were abandoning before the page finished loading.
Paid advertising cost per lead improved over the subsequent quarter. Better landing page quality scores, driven in part by improved user experience signals, reduced average CPC across the campaigns pointing to the rebuilt pages.
"The page felt faster and more trustworthy, even though the core messaging stayed the same."
What this means for service marketing sites
The takeaway from this project is not that every service site needs a full rebuild. It is that performance bottlenecks on the pages that drive enquiries have a measurable cost — both in conversion rate and in advertising efficiency — and that targeted performance work on those specific pages can deliver a strong return.
The priority order: audit the pages that send the most paid traffic first. If those pages have poor Core Web Vitals on mobile, that is the highest-value performance investment available.
The changes required are often not design changes. In this case, the page looked identical after the rebuild. The work was in how the page loaded, not in what it showed.
Recommended next step
Measure the pages that send the most leads, identify which have poor LCP or CLS scores on mobile, and apply the same performance principles to those pages first. The combination of hero image optimisation and third-party script deferral addresses the most common causes of poor mobile LCP on service marketing sites.