Structure recommendation
Subfolders, subdomains or country-code domains, with the authority and operational trade-offs stated plainly rather than a default asserted as best practice.
International SEO decides how a site serves several countries or languages without those versions competing with each other. Most of the difficulty sits in two places: choosing the URL structure before anything is built, and implementing hreflang correctly — which fails silently when it is wrong.
Typically a fixed-scope audit, then ongoing market support.
First, URL structure: country-code domains, subdomains or subfolders. Second, whether you are targeting languages, countries, or both — because Spanish for Spain and Spanish for Mexico are different targets. Both decisions are cheap before a build and expensive to change once markets are live.
Structure is largely settled in practice. Subfolders consolidate authority across markets and are the right default for most organisations. Country-code domains give the strongest local signal and split your authority into separate sites, which suits businesses with genuinely independent operations per market and punishes everyone else.
The language-versus-market distinction is where most sites go wrong conceptually. Publishing one Spanish version for every Spanish-speaking country means competing against yourself in each, with currency, spelling and legal differences unaddressed. Whether that matters depends on how different your commercial offer actually is between those markets.
Hreflang is where implementations fail, and the failure mode is what makes it dangerous. Wrong hreflang does not throw an error or produce a warning anyone reads — it quietly serves the wrong version to the wrong market, and the symptom is a market underperforming for reasons nobody can identify. Return-tag errors are the most common cause.
The fourth issue is content that was translated rather than localised. Machine translation of a page written for one market produces text that is technically correct and commercially wrong: the wrong examples, the wrong objections, the wrong regulatory framing. It ranks poorly because it genuinely serves that market poorly.
The engagement covers a structure recommendation with the trade-offs stated plainly, full hreflang implementation and validation including return tags, market-versus-language targeting decisions, localisation guidance that goes beyond translation, and per-market reporting so each version is judged on its own performance.
Subfolders, subdomains or country-code domains, with the authority and operational trade-offs stated plainly rather than a default asserted as best practice.
Complete annotation including return tags and self-references, implemented where it can be maintained rather than hand-written into pages that will drift.
Verification that the annotation actually works, since incorrect hreflang produces no error and its symptom is a market quietly underperforming.
Whether each version targets a language, a country or both, decided against how different your commercial offer genuinely is between those markets.
What must change beyond translation — examples, currency, regulatory framing, objections — so a market version serves that market rather than reading as imported.
Each version measured separately, because an aggregate hides the one market where hreflang is misdirecting traffic and the problem stays invisible.
The engagement audits your current structure and hreflang first, since errors there invalidate everything downstream. It then settles the market-versus-language targeting, implements and validates the annotation properly, and establishes per-market reporting so each version is judged on its own terms.
What exists, whether the annotation is valid, and whether return tags resolve. This regularly finds markets that have been mis-served for years unnoticed.
→ A structure and hreflang validity report
Language, country or both, per version. Decided against how different your commercial offer genuinely is rather than by defaulting to one version per language.
→ An agreed targeting matrix
Generated from a single source so it cannot drift, with return tags and self-references complete. Hand-maintained hreflang breaks the first time a page moves.
→ Generated, complete hreflang annotation
Verification against live pages, because incorrect hreflang throws no error and the only symptom is a market performing worse than it should for no visible reason.
→ A validated annotation report
Each version tracked separately. Aggregate international reporting reliably conceals the single market where something is wrong, which is the one you needed to see.
→ Per-market performance reporting
This decision is cheap before a build and expensive afterwards, so it is worth making deliberately. The trade-off is between consolidating authority across markets and sending the strongest possible local signal — and most organisations overestimate how much they need the latter.
| Structure | Authority | Best suited to |
|---|---|---|
| Subfolders (/de/) | Consolidated across all markets | Most organisations, most of the time |
| Subdomains (de.) | Partially separated | Separate infrastructure or platform per market |
| Country domains (.de) | Fully separated per market | Genuinely independent operations per country |
You need this when the wrong country version ranks in a market, when a market underperforms for no identifiable reason, when hreflang was hand-written into pages, or when you are about to launch a new market and the structure has not been decided yet.
Correcting hreflang typically shows within weeks, because it changes which version engines serve rather than requiring new authority. Localisation gains take longer and matter more, since a version that genuinely serves its market outperforms a translated one regardless of annotation.
Each answer is written to stand alone in 40 to 60 words — the shape an AI Overview or Perplexity citation lifts. Ships with FAQPage schema.
Subfolders for most organisations, because authority consolidates across markets rather than being split. Country-code domains give the strongest local signal and separate your authority entirely, which suits businesses with genuinely independent operations per market and penalises everyone else.
Very frequently hreflang. Incorrect annotation produces no error and no warning — it quietly serves the wrong version to that market. Return-tag errors are the most common cause and are invisible unless somebody validates the implementation against live pages.
It depends on how different your commercial offer actually is. One Spanish version for every Spanish-speaking market means competing against yourself with the wrong currency and regulatory framing. Where the offer is genuinely identical, one language version is fine.
For comprehension frequently yes; for ranking rarely. Translated content keeps the original market's examples, objections and framing, so it serves the new market poorly and performs accordingly. Localisation changes what the page argues, not only what language it argues in.
Validation against live pages, checking that return tags resolve and self-references exist. This is not optional — incorrect hreflang is silent, so the only way to know is to verify rather than to assume it was implemented correctly.
With subfolders, straightforwardly. With country-code domains, each new market is effectively a new site needing its own authority built from nothing. That asymmetry is the strongest practical argument for deciding structure carefully before the first market launches.
Thirty minutes with a senior strategist. We pull your live visibility while we talk and tell you plainly whether a international SEO is what you need — or whether your problem sits somewhere else.