The first decision is the one everyone gets wrong
ccTLDs, subdomains or subfolders. It is a ten-minute conversation and a five-year consequence.
| Approach | Best when | The cost |
|---|---|---|
| Subfolders example.com/de/ | You want one domain accumulating all authority, and you are entering markets gradually. Our default recommendation for most businesses. | Weaker local signal than a ccTLD, and one platform has to serve every market. |
| ccTLDs example.de | You have real local operations, local teams, and the budget to build authority separately in each market. | You start from zero in every country. Expensive, slow, and often chosen for emotional reasons. |
| Subdomains de.example.com | Technical constraints force separation, or each market runs its own stack. | Authority sharing is unreliable and you will maintain the same fixes several times. |
Hreflang, done so it actually works
Hreflang is simple in principle and broken on most sites that implement it. These are the five faults we find every time.
- A generated tag set, validated automatically, so it cannot drift as pages are added
- x-default set deliberately for visitors we have no market for
- Per-market Search Console verification, so you can actually see which market is failing
- Missing return tags — every alternate must point back, or the cluster is ignored
- Language codes invented rather than looked up (
en-ukinstead ofen-GB) - Tags pointing at redirecting or canonicalised URLs
- A self-referencing tag left out of the set
- Hreflang and canonical contradicting each other on the same page
Localisation is not translation
Translating a page keeps the words and loses the demand. People in different markets search for the same product using different terms, in different formats, with different objections — and sometimes the product itself is not the one they want.
So market research comes before translation. We run keyword research natively in each market, which routinely turns up that the head term you translated has a tenth of the volume of the phrase locals actually use.
Then a native speaker writes, rather than edits. Machine-translated commercial pages read as machine-translated, and buyers notice faster than search engines do.
- Native keyword research per market, not translated keyword lists
- Currency, units, date formats, phone formats and address conventions
- Local proof: local case studies, local reviews, local trust marks
- Legal and compliance copy checked per market
- Local hosting or CDN edge coverage where latency affects conversion
International FAQs
Which market should we enter first?
The one where you can already service customers well. We model search demand, competition and existing referral traffic to rank the options, but operational readiness beats search volume every time — ranking in a market you cannot deliver to is an expensive way to annoy people.
Do we need separate content for the US and the UK?
Usually yes, and less for the spelling than the substance: different regulations, different competitors, different price expectations. At minimum, separate the commercial pages and share the informational ones.
Can you handle non-English markets?
We handle strategy, technical implementation and QA in-house, and work with native writers and reviewers per market. We will not pretend to have in-house fluency we do not have.