E-commerce performance rescue

The problem

Describe the symptom as the client experienced it, and what had already been tried.

Approach

• Field data from the Chrome UX Report alongside lab profiling, to separate a slow server from a slow page.
• Time-to-first-byte measured first. It identified page generation, not the frontend, as the dominant cost.
• Slow query log analysis surfaced missing indexes on the largest tables and an autoloaded options table that had grown unchecked.
• Plugin asset audit: scripts and styles being queued site-wide were restricted to the pages that actually used them.
• Hero imagery moved to AVIF with correct fallbacks and preloading, and dimensions added everywhere to eliminate layout shift.