Measure First, Then Act
Get real numbers before changing anything. We look at three indicators: time to first byte, largest contentful paint, and interaction responsiveness. Run both lab data and real-user data, because a page that feels instant on a development machine can behave very differently on an overseas network.
Priority One: Images
Images are usually the heaviest burden on an export site. Three moves solve most of the problem:
- Convert everything to WebP, roughly 30% lighter than JPEG
- Export at the displayed size — no 4000-pixel uploads shrunk by CSS
- Lazy-load everything below the first screen
We saw one client site drop from 4.2 seconds to 2.3 seconds on those three changes alone, without touching a line of layout code.
Priority Two: Render-Blocking Resources
CSS and JavaScript block painting. The fix is straightforward: inline critical styles, defer the rest, add deferred attributes to scripts, and load third-party analytics or chat widgets last.
Watch chat widgets in particular. Many export sites load them at the very top of the page, so the browser connects to three overseas servers before the buyer sees a single product.
Priority Three: Delivery and Caching
Enable compression, set long cache lifetimes for static assets, and use nearby edge nodes. If your main customers are in Europe and the server is in Asia, network round-trips alone can cost hundreds of milliseconds. This step usually needs to be handled together with DNS and CDN configuration.
Priority Four: Fonts
Custom fonts are common and easy to overlook as a performance cost. Limit the number of weights, preload the primary font, and set a system-font fallback.
Optimisation Is Ongoing
Passing on launch day does not mean passing six months later. Every large image the content team adds and every plugin operations installs can erase the gains. Put a speed check into the monthly review rather than doing it once at launch.



