The moment you sell in a second country, SEO stops being about one site and starts being about how several sites behave as a system. Architecture, hreflang, content parity and per-market intent — four problems that do not exist in single-market SEO and that no amount of domestic best practice will solve.
In my day job I look after 21 web properties across nine countries — consumer health portals, professional platforms and brand sites, from 37 URLs to 3,100 — with technical audits running weekly on all of them. The same four failures come up again and again, in every industry I have seen them in.
Subfolder, subdomain or country domain is usually decided by whoever set up the CMS, not by anyone weighing the trade-off. For almost every company below enterprise scale the answer is subfolders on one domain: they inherit authority you have already earned, one technical fix applies everywhere, and the maintenance cost stays flat as you add markets.
Country domains earn their keep when you have separate legal entities, genuinely different products per market, or a brand name you cannot use somewhere. Subdomains are usually the worst of both worlds — the overhead of separate sites without the local trust signal of a country domain. It is a hard decision to reverse, which is why it deserves analysis rather than a default.
Hreflang fails silently. There is no warning, no penalty — the annotations are simply discarded and your markets compete with each other. The errors I find most often, in rough order of frequency:
en-UK; the country code is GB. Region codes must be ISO 3166-1 Alpha-2, language codes ISO 639-1.x-default, so Google has nowhere to send everyone you did not explicitly target.The most expensive problem, and the least visible. A second market launches with a translated subset of the first market's pages: fewer pages, thinner ones, no local links. Then everyone concludes that Google distrusts the new market. It does not — there is simply less there to rank.
The fix is rarely more translation. It is working out which pages actually earn traffic in that market's SERP, and building those properly.
Your German buyers do not search using translated versions of your English keywords. They use different words, arrive with different objections, and compare you against a different competitive set. A page that was translated rather than localised reads correctly and ranks for nothing. Per-market keyword and intent mapping is the work; translation is a downstream task.
A full crawl per market, indexation and canonical review, hreflang validation across every pairing, and a parity matrix showing which pages exist where. This is the part most teams have never seen laid out.
Keyword and intent research done natively in each language, not translated. Volume, difficulty and commercial intent per market, with the competitor set that actually ranks there — which is rarely the one you assume.
A written recommendation with the trade-offs stated, including what it costs to migrate if you are already on the wrong structure, so the decision can survive a conversation with your CTO.
A technical specification your developers can implement without interpreting it, and a content plan your writers can execute without guessing. Prioritised by impact and effort, with an owner against each item.
Per-market segmentation in Search Console and analytics, plus rank and AI-citation tracking, so you can tell in month three whether it is working rather than in month twelve.
Not theory. Nine markets, in healthcare — a category where YMYL scrutiny, medical review workflows and regulated claims make every content decision slower and every technical mistake more expensive. If it works there, it works on your SaaS site.
en-UK — the country code for the United Kingdom is GB.