Same storefront, less for the browser to do
This was not a redesign. 2XU's storefront looked the way the brand wanted it to look. The question was how much work the browser had to do to show it, on a phone, on a regional store, with every app and campaign the theme had accumulated.
That meant working on the assets sent to the browser, the order they load in and the shopping interactions they support, across regional storefronts that each have their own configuration and needed a repeatable way to receive changes.
Preparing the assets and the build
We reviewed the theme's assets, removed files no longer in use, took a passive video player and its media loading off the page, and updated how fonts are delivered. The JavaScript build moved to Vite while keeping the existing Preact components, and stylesheets joined a repeatable build that left the compiled files about 20% smaller, with Tailwind keeping its own step. Compiled files kept the names the theme's Liquid templates already use, so the source became easier to maintain without disturbing the storefront's structure.
What loads first
The first thing a shopper sees gets priority. Image preloads follow the mobile or desktop layout, and images further down the page wait their turn. Search and product features load their assets where they are used rather than everywhere, individual widgets need less data, and link prefetching prepares the pages a shopper is likely to open next. Shared loading controls let third party features be timed around the initial page load, while each regional store keeps its own app configuration.
Interactions that stay responsive
Product rows use native scrolling. Search results appear as soon as product data arrives, with swatch details added afterwards and the shopper's selected colour preserved. The interactions stay responsive while supporting information is still on its way.
Releasing to every store
Release tooling checks each store's settings, records the result of each deployment and identifies the revision that is live, with a documented rollback process. The related staging tool is connected to ten of 2XU's regional stores and prepares staging themes for the ecommerce team to review rather than publishing them. Between 4 August and 12 September 2026 it recorded 80 successful staging releases across nine stores through 21 release workflows.
The results, and what each one measures
In controlled mobile Lighthouse audits on preview themes, LCP fell by 89.2% on the audited product listing page and 91.3% on the audited product page. In a separate controlled test on the New Zealand homepage preview, JavaScript transfer fell 40.2%, from about 518 KB to about 310 KB, total page transfer fell 17.8%, and the page made eight fewer requests, each figure the median of three runs. These are preview lab results, not live shopper performance, and they are not conversion results.
Live field data tells its own story. Google's public Chrome UX Report for the 2xu.com homepage on mobile shows LCP of 1.61 seconds in the latest 28 day window, 9 August to 5 September 2026, with INP of 182 milliseconds and CLS of 0.00, all three Core Web Vitals rated good and stable. That is 69% below the January 2026 peak of 5.19 seconds and 57% below the earliest public window, November to December 2025.
The field trend cannot be pinned to individual releases, so we present it as the public record of the storefront over the period of the work rather than as a controlled measurement of it, and we keep the preview and field figures apart.
A smaller build and clearer priorities
The storefront kept its design and content workflow, with more selective feature loading and clearer priorities for the browser.
For the developers who work on it next, source changes go through repeatable builds and checked releases to each store, with an identifiable revision and a documented way back.