The honest answer is: it depends on your specific situation, and we give you a definitive recommendation with full rationale after an architecture assessment. In general terms: ccTLDs send the strongest geographic signals and are most trusted by local users, but require building domain authority independently for each country — meaning you're starting from zero in every market. Subdirectories inherit the root domain's authority and are the easiest to manage — but geographic signals are weaker than ccTLDs. Subdomains sit in the middle on both dimensions. The right answer for your business depends on how many markets you're targeting, your current domain authority, your content volume, your development resource, and your expansion timeline. We make this recommendation based on your actual situation — not a generic default preference.

Hreflang is a technical tag that tells Google which version of a page to show to users in which country and language. It requires a confirmed bidirectional relationship between every page and every language alternate — so if you have a UK page, a US page, and a German page, all three need to reference each other correctly in both directions. The reason it keeps going wrong is that hreflang is deceptively complex to implement at scale. Most implementations are done by developers following a blog post, without fully understanding the edge cases: what happens with pagination, with dynamically generated pages, with auto-inserted canonical tags, or with incorrect country-language code combinations. A single error pattern at scale can produce hundreds or thousands of broken tag relationships — each one silently diluting rankings across all affected markets.

Fixing hreflang errors typically produces measurable ranking improvement in the home market within 4–8 weeks, as Google re-evaluates the correct page relationships and consolidates authority to the right variants. Rankings in new markets with correct hreflang implementation take longer — typically 3–6 months for meaningful organic traffic, as Google needs to crawl, evaluate, and rank the new country variant content. Market-specific content built around local search intent compounds significantly faster than translated content — because it targets actual local queries rather than translated versions of English keywords that local users don't search for.

It is never too late to fix international SEO architecture. The earlier the better — every month with a broken hreflang implementation accumulates ranking damage, dilutes home market authority, and makes the fix progressively more complex as more content is added to the broken structure. But the fix is almost always worth doing regardless of how long the wrong architecture has been in place. For businesses considering a full architecture migration — moving from subdomain to subdirectory, for example — we recommend a careful, staged approach. A poorly executed migration can cause short-term ranking disruption even when it is ultimately the right move. We specify the migration approach in detail as part of the engagement, including exactly what to do and in what sequence to minimise disruption.

Yes — significantly so for markets where Google is not dominant. Baidu (China's primary search engine) has completely different ranking factors: simplified Chinese content, ICP licensing for China-hosted sites, Baidu-specific structured data, and a strong preference for content hosted on Chinese servers. Yandex (Russia) similarly rewards content hosted in Russia and written specifically for Russian search behaviour. Naver (South Korea) operates as much as a content platform as a search engine — with its own blog, Q&A, and shopping ecosystems that require separate optimisation strategies. We identify which search engines matter for each of your target markets and produce market-specific technical and content requirements. A Google-only international SEO strategy will consistently underperform in markets where Google isn't the primary discovery channel.

Because translation and localisation are fundamentally different things. Translation converts your words into another language. Localisation adapts your content to serve the actual search intent that exists in that market — which is often significantly different from the English original. A UK software company's content might translate accurately into German, but the terms German buyers search for when researching that type of software are often completely different from translated English keywords. Local buyers use different phrases, ask questions differently, and research in ways shaped by local business culture. Localised content starts with in-language keyword research conducted from scratch in that market — and builds the content around what local users actually search for. Content that begins as English and gets translated consistently underperforms content written for local intent from the start, regardless of translation quality.