Test the performance change in isolation
Rakuten 24 used a month-long A/B test to compare a Core Web Vitals optimized landing page with its original version. The test split traffic 50/50, and the versions had no other functional or visual differences [1]. web.dev reports relative increases of 33.13% in conversion rate, 53.37% in revenue per visitor, and 15.20% in average order value [1]. These are relative changes within that test, not percentage-point increases.
That separation gives the result a useful interpretation: the test evaluated a performance-focused change without a redesign confound. The source does not provide a basis for claiming the same result on another site, nor does it establish that every performance project will improve all three business measures.
Read conversion and order value together
The reported conversion-rate increase describes a larger relative rate of visits becoming conversions; revenue per visitor captures value across visits; average order value describes the size of completed orders. Our interpretation is that these measures answer different business questions. A conversion gain can coexist with a smaller or larger order-value change, so teams should review both the number of converting visits and the value of those orders.
The source reports the three relative test outcomes but supplies no absolute baseline values in the published case [1]. It would therefore be misleading to calculate an absolute revenue gain from the percentages alone. A team applying this approach should first capture its own baseline counts and revenue definitions, then select a primary outcome before testing.
Prioritize likely sources of page friction
As a proposed diagnostic sequence, inspect large images for appropriate dimensions and formats, reserve layout space so content does not jump as media loads, and review whether JavaScript delays important content or interactions. These are practical areas to investigate, not claims about additional Rakuten implementation details. The source documents the test and its outcomes, but this study does not add project facts beyond those reported.
The order is deliberate: images can dominate what appears first, reserved space can reduce disruptive movement, and unnecessary or late JavaScript can hold up rendering or response. Core Web Vitals guidance describes loading, interactivity, and visual stability as distinct dimensions of user experience [2]. A measured bottleneck should determine which intervention is tested first.
Build guardrails into content publishing
A suggested CMS workflow would give authors image size guidance, preserve explicit dimensions or aspect ratios where appropriate, and preview pages with representative content before publication. A release checklist could flag unusually large media, missing dimensions, or scripts added to high-traffic templates. These are proposed authoring controls, not reported features of Rakuten 24's CMS.
For regression checks, compare representative page templates in a lab before release and monitor field measurements after deployment. Core Web Vitals guidance recommends field measurement for real user experience and notes that lab checks help catch regressions during development [2]. The two views serve complementary purposes: a controlled check can point to a change, while field data reveals how it behaves across users and conditions.
Set a local test and interpret its limits
A practical next step for another commerce team is to choose one high-traffic landing page, preserve a control, and isolate a defined performance change. Specify the primary outcome, such as conversion rate, and track revenue per visitor and average order value as separate supporting outcomes. Predefine the test window and verify traffic assignment and event tracking. Those are proposed test steps, not details attributed to Rakuten.
The case reports a successful test for Rakuten 24 and shows that performance can be assessed alongside commercial outcomes [1]. It does not prove a universal causal effect across sites, reveal absolute commercial values here, or establish that a particular optimization caused each metric's change independently. A local test can establish whether the measured result repeats under a site's own traffic and purchase patterns.
Sources & useful references
Results are attributed to the original publisher. Proposed tests and creative recommendations are our analysis.
