- Decide the structure first: subdirectories, subdomains or separate domains. Changing it later is a migration.
- Spanish for Spain and Spanish for Latin America are different markets, not the same page with a flag.
- hreflang is fiddly and unforgiving. Get it right or leave it out.
- Translation is not localisation. Prices, examples, regulations and search terms all differ.
A Madrid business expanding into other markets faces a set of decisions that are cheap now and expensive in eighteen months. These are the ones that matter.
Structure: three options, one recommendation
Subdirectories (example.com/mx/) are the default recommendation: they inherit the authority of the main domain, they are cheap to maintain, and one technical setup serves everything. The drawback is that everything shares a fate - a penalty or a technical problem affects all markets.
Subdomains (mx.example.com) separate concerns more and are sometimes required by infrastructure, but authority transfers less cleanly.
Country domains (example.mx) give the strongest local signal and the clearest legal separation, and they cost the most: each one starts from zero authority and needs its own programme.
For most businesses leaving Spain for the first time, subdirectories with correct hreflang is the right answer.
Language versus market
| Structure | Example | Strength | Cost |
|---|---|---|---|
| Subdirectories | site.com/fr/ | Inherits the authority you already have | Harder to run as a separate business |
| Subdomains | fr.site.com | Clean separation, separate hosting | Authority transfers less readily |
| Country domains | site.fr | Strongest local signal, clearest to users | Each one starts from nothing |
For a company expanding from Madrid rather than opening a genuinely separate business, subdirectories are almost always the right answer. The other two are for companies with local entities, local teams and the budget to build authority three times.
Spanish speakers in Mexico, Argentina and Spain search differently: different vocabulary for the same product, different price expectations, different competitors, different seasons. A single Spanish page targeting all of them will underperform in each.
Where the offer is genuinely identical, one page with hreflang variants by region is defensible. Where prices, delivery or regulations differ, the pages have to differ too, or the conversion rate does the complaining.
hreflang, carefully
Language and market are not the same axis, and conflating them is the most common strategic error in an expansion:
| You are targeting | Code | Watch out for |
|---|---|---|
| Spanish speakers anywhere | es | Prices and terms that only apply in Spain |
| Spain specifically | es-ES | Missing a plain "es" fallback for the rest |
| Mexico specifically | es-MX | Vocabulary, currency and delivery differences |
| English speakers anywhere | en | Nothing - a sound default for a first expansion |
| Anyone not matched | x-default | Forgetting it, which leaves the choice to guesswork |
- Every page in a set references every other, including itself.
- Use language plus region where the distinction matters: es-ES, es-MX, en-GB.
- Include an x-default for visitors outside the set.
- Keep the annotations consistent between the sitemap and the head, and validate them after every deployment.
Wrong hreflang is worse than none: it sends the wrong market to the wrong page and hides the problem behind plausible-looking markup.
Localisation beyond words
The commonest half-expansion. The page reads correctly and asks the visitor to do something impossible.
If A points to B and B does not point back, the pair is discarded. It must be reciprocal and self-referencing.
It traps crawlers on one version and infuriates travellers. Suggest, do not force.
Currency, tax display, payment methods, phone formats, addresses, delivery expectations, and the examples in the copy. A page for Mexico that quotes euros and references Spanish regulations reads as an afterthought whatever the translation quality.
Keyword research has to be redone per market too. The literal translation of your best Spanish term is frequently not what people in the target market type.
Sequencing an expansion
A month of paid search in the target market tells you more than any keyword tool, and costs less than a translation.
Full localisation, local proof, a working route to enquiry. A half-built market teaches you nothing.
Its own Search Console property and its own conversion tracking, or you will never see it working.
The mistakes from the first market are cheap to fix once and expensive to replicate three times.
One market at a time. Research demand before building anything, launch with the core commercial pages rather than the whole site, measure for a quarter, and only then decide whether to deepen or to move on. Businesses that launch six markets simultaneously usually end up maintaining six half-sites.
Frequently asked questions
Do we need a local address to rank in another country?
Not for organic, though it helps for local results. Structure, language targeting and local links carry most of the weight.
Is automatic translation acceptable?
As a first draft, edited by someone who knows the market. Published raw, it reads badly and converts worse, whatever it does to rankings.
Should we redirect visitors by IP?
Which market should we open first?
The one where you already receive enquiries you currently turn away. Existing accidental demand is the cheapest evidence there is.
Do we need separate content or can we translate?
Translate the structure, rewrite the proof. Examples, prices, regulation and objections are what make a page local; the argument underneath usually travels.
Prefer a suggestion banner over a forced redirect. Automatic redirects block crawlers from seeing the other versions and frustrate travelling users.
Planning an expansion?
We size the demand market by market and set the structure up once, so it does not need redoing.
Talk to Madrid SEORelated: SEO services.