<b>The language-vs-region trap: when es-MX and es-ES fight for the same query</b>

<b>The language-vs-region trap: when es-MX and es-ES fight for the same query</b>

Hypothesis: over-specifying region codes (es-MX, es-AR, es-CO) when your content is functionally identical Spanish creates more cannibalization risk than it solves.

We tested this on a catalog estate with 11 Spanish regional variants sharing ~95% identical copy.

Findings:

— For queries with no regional intent, Google frequently selected an 'unexpected' variant — es-CO ranking in Spain — because hreflang only influences which URL is swapped in, not which one is deemed most relevant in the first place.

— Regional variants that differed only in currency and a phone number were treated as near-duplicates. The annotation prevented hard duplicate filtering, but consolidated authority did not flow as cleanly as a single canonical would have.

— Where genuine localization existed (local payment methods, regional product availability, idiomatic copy), the regional split paid off — correct variant served, lower bounce.

Nuance: language-only targeting (es, no region) is underrated. It sidesteps the reciprocity overhead of N regional pages and works well when your differentiation is linguistic, not commercial.

Limitation: we can't measure the counterfactual revenue of consolidation versus split on the same estate simultaneously.

Conclusion: split by region only where the content genuinely diverges. Otherwise prefer language-level targeting. Granularity you can't justify with real differences is overhead that dilutes signals.