Performance Optimisation at Scale

The challenge

Global brand clients came in with production platforms that looked polished but loaded slowly — heavy bundles, unoptimised images, and chained network requests pushing LCP past 4 seconds and hurting conversion.

The approach

• Profiled LCP and network waterfalls first — measurement before optimisation, so every change targeted a real bottleneck.
• Code-split and lazy-loaded modules, optimised and right-sized images, and trimmed API payloads.
• Layered in smart caching and CDN delivery, plus deliberate loading and error states.
• Turned the process into a repeatable playbook applied across multiple client platforms.

The outcome

• LCP cut from 4.2s to 1.8s (-57%).
• Bundle size reduced by ~35%.
• The same discipline now applies to everything I ship — performance budgets from day one, not as a rescue.

Stack