x-default pointing to a language selector vs to English: A/B

x-default pointing to a language selector vs to English: A/B

Hypothesis: pointing x-default at a neutral language-selector page would serve unmatched-locale users better than pointing it at the English homepage. Methodology: one global product, we split ~600 entry-point URLs into two groups — group A's x-default resolved to a /select-region/ page, group B's to /en/. 8 weeks, measuring bounce, language-correct landings, and downstream conversion.

Findings:
— Group A (selector) had a 28% higher bounce on the landing step — users hit a friction page. Expected.
— But group A's eventual language-correct deep-page rate was 14 points higher: people who self-selected ended up on the right locale.
— Conversion told a messier story: group B (English default) converted 11% better overall, because the bulk of unmatched traffic was English-comfortable anyway and the selector added a step.
— For genuinely non-English unmatched traffic (a ~9% slice we could segment), the selector won — conversion +22% in that slice.

Caveats: x-default is a hint to search engines about which URL to show for unmatched locales, not a redirect — so part of what we measured is landing-page UX, not pure hreflang behavior. The two are entangled here.

Conclusion: English-default wins on aggregate conversion; a selector wins only if your unmatched traffic is genuinely multilingual. Know your unmatched mix before choosing.