What happened when we added x-default to a 14-market SaaS

What happened when we added x-default to a 14-market SaaS

Hypothesis: a missing x-default was bleeding US organic traffic into geo-IP redirect loops. Methodology note: we instrumented one property, 14 language-region pairs, ~2,100 indexed URLs, 90 days pre/post, controlling for two algorithm updates by indexing the control subfolder that we left untouched.

What we found, broken out:
— Before, the bare /product/ URL had no return tag pointing anywhere; Google was serving the en-gb variant to US searchers 38% of the time (measured via GSC query-country splits).
— Adding a self-referencing x-default on the generic English page cut mismatched-country impressions on those queries from 38% to 9% over six weeks.
— Clicks to the correct US variant rose 22%; total clicks rose only 11%, so roughly half the gain was redistribution, not net new demand.

Nuance and caveats: the redirect loop itself was an aggravating factor, so we can't cleanly separate the x-default effect from the redirect fix shipped the same week. The 22% figure is therefore an upper bound. We also can't rule out seasonality on two of the fourteen markets.

Conclusion: the data suggests x-default matters most as a tiebreaker when geo-IP redirects already muddy the signal — not as a standalone lever. Single property, so treat the magnitude as directional.