International SEO
ccTLD versus subdomain versus subfolder is international SEO's structure decision — where each market's version lives — and it's the general consolidation physics with a second force added: country signalling. The trade is clean once stated: subfolders pool authority and signal country weakest; ccTLDs signal country strongest and pool nothing; subdomains sit in the ambiguous middle — and the decision follows from which force your business actually needs. Here's the full comparison and the decision tree.
The three structures, traded
- Subfolders (example.com/de/): every market's content feeds one domain's authority pool — a link earned anywhere strengthens everywhere, each new market launches with the domain's accumulated trust instead of a zero start, and operations stay singular (one platform, one governance stack). Country signal: none inherent — supplied by hreflang, Search Console geotargeting per folder, and on-page localisation. The default recommendation for most businesses, and the reason is arithmetic: authority is the scarce resource, and subfolders waste none of it.
- ccTLDs (example.de): the unambiguous country flag — strongest geo-signal to engines and users (local-TLD trust in markets like Germany is a real CTR and conversion effect the physics discussions under-weight), full local identity, sometimes legal/hosting fit. Costs: authority split completely (every domain earns its own links forever — the multi-market link programme priced per the sequencing doctrine), operations multiplied, and defensive registration across markets. The right call when country identity is commercially decisive and the per-market investment is real and permanent.
- Subdomains (de.example.com): the middle nobody quite recommends: partial-at-best consolidation (the general guide's ambiguity), weak-ish country signal (geotargetable like folders), separate-infrastructure flexibility as the one genuine advantage. Chosen mostly by platform constraint; migrate-to-folders remains the standing advice when feasible.
The decision tree
Default: subfolders — unless a specific force overrides: ccTLDs where local-domain trust demonstrably moves your market's buyers (consumer commerce in strong-TLD cultures), where legal/data residency demands it, or where genuinely independent country operations (teams, catalogues, P&L) make the split real anyway; subdomains where platform architecture forces them (accepted, not chosen); and hybrids are legitimate — a ccTLD for the flagship market whose trust economics justify it, subfolders for the rest, is a common grown-up answer. Whatever the choice: made once, per the migration-debt pricing — structure changes at international scale are migrations multiplied by markets.
Frequently asked questions
Does a .com with subfolders really rank in Germany against .de sites?
Routinely — geotargeted, localised, locally-linked subfolder versions compete fully in local results; the ccTLD's ranking edge as a pure signal is modest and swamped by content and authority quality. The .de advantage that persists is the user-side one: SERP trust and click-through in TLD-conscious markets — measure whether your market has it before paying the split-authority price for it.
We already have ccTLDs. Consolidate to subfolders?
The migration math per market: consolidation pools authority permanently at the cost of the settling window and any local-trust effects — worth it for thinly-resourced ccTLD sprawls (five weak domains becoming one strong one is the classic win), rarely worth it for established ccTLDs with real local equity. The checklist per domain migrated, exact-counterpart maps included.
Do generic new TLDs (.shop, .berlin) geo-signal anything?
Google treats most new TLDs generically (geotargetable like .com); the city TLDs carry no ranking geography either — branding decisions, not signals, per the TLD rules. The structures above remain the whole geo-toolkit — wired with hreflang, filled per market, and ranked by the usual pair: localised content and authority earned in each market's web (our markets list is long).