Verified as of August 2026: EU and UK rules are actively changing, and some widely-shared claims are out of date. This is operational orientation, not legal advice; confirm with official sources or your data protection adviser.
Hreflang is a solved problem with a thousand tutorials. The reason multilingual SEO programmes in Europe underperform is almost never the annotation layer: it's that the keyword set was translated rather than researched, the markets were entered in the wrong order, and the link acquisition strategy assumed one European link ecosystem where there are at least six. If you get per-language keyword research, market sequencing and per-market authority right, a technically imperfect hreflang implementation still wins. The reverse is not true.
Key Takeaways
- Translated keywords rank for the wrong intent constantly. Germans search compound nouns; French searchers switch register; Nordic professionals mix English terms into native queries.
- Search volume in a language is not search volume in a market. German queries come from Germany, Austria and Switzerland with different competitive dynamics and different SERPs.
- Sequence matters: DACH first for volume and B2B spend, then Nordics for English-tolerance and fast wins, then Benelux, then Southern Europe where the linguistic and link work is heaviest.
- Every market has its own link ecosystem: trade associations, regional press, professional bodies. A German site earns German links or it doesn't rank in Germany.
- Machine translation plus human editing is viable. Machine translation alone is a ranking ceiling and a trust ceiling.
- The AEO layer changes the calculus: answer engines cite native-language sources differently, and native-language depth is now an acquisition channel, not just a localisation cost.
A Note on Where This Comes From
I do most of my work from India, which is not a monolingual search market by any definition. Running organic growth for an edtech brand across Hindi, English and a set of regional-language audiences: the programme that took one client from 26,000 to 117,000 on Instagram and 50,000 to 160,000 on LinkedIn, forces you to learn the same lessons Europe teaches: that a keyword translated is a keyword mis-specified, that formality register is a ranking variable, that code-mixing is normal user behaviour rather than an edge case, and that authority is earned inside a language community rather than transferred into it.
The specific languages differ. The structural problems are the same, and I'd rather say that plainly than pretend European multilingual SEO is exotic.
Why Translated Keyword Lists Fail
The intent drift problem
Take a English seed term like "marketing automation software." Translate it to German and you get something plausible. Research German instead and you find the highest-intent commercial queries cluster around different compounds, that "Software" is sometimes dropped entirely because the compound implies it, and that a meaningful share of the German B2B audience searches the English term anyway because the category vocabulary arrived in English.
A translated list gets you a page targeting a term with volume that nobody with buying intent actually types.
German: compounds are the whole game
German forms compound nouns freely, and the compound is frequently the search term. Keyword tools split them inconsistently. Two consequences:
- Your keyword research must work with the compound as a unit, and you should check singular, plural and the common hyphenated variant separately.
- Content structure follows: German pages often need to address the compound concept directly in the H1 rather than approaching it through a phrase.
Austrian and Swiss German add vocabulary differences (and Swiss German drops the ß). If DACH matters commercially, treat Switzerland as a distinct market rather than a German page with a currency switch.
French: register is a ranking variable
French has a real formal/informal distinction (vous/tu) and a stronger convention around anglicism avoidance in professional contexts, reinforced institutionally. A French B2B page written in the register of a US SaaS blog reads as unserious to the audience it's meant to convert.
Practically: research the French terms actually used in the relevant professional community, not the ones a translation tool prefers, and pick a register and hold it. Mixed register is worse than either choice.
Nordics: English mixing is not a bug
Swedish, Danish, Norwegian and Finnish professional audiences code-switch heavily. Many technical and business queries are typed in English even by native speakers, because the terminology is English. Others are typed natively.
This means Nordic markets often have a genuinely lower localisation cost, a good English page can rank and convert, but it also means the queries that are native-language are high-intent and under-contested. The Nordic play is usually English for the category pages and native language for the specific, bottom-of-funnel and locally-regulated topics.
Southern Europe: volume in the long tail
Spanish and Italian search behaviour trends toward longer, more conversational queries. Spanish also splits across Spain and Latin America, and you need es-ES targeting discipline or you will attract traffic you cannot serve.
Dutch: the English shadow
The Netherlands has very high English proficiency and Dutch searchers frequently accept English results. But Dutch-language queries convert better for anything involving contracts, compliance, pricing and procurement. The pattern: English for discovery, Dutch for decision.
Market Sequencing: The Order That Actually Works
Most teams enter markets in the order their sales team asks for. Here's the order that compounds.
1. DACH
Germany, Austria, Switzerland. Largest European B2B software and services spend, high search volume, strong purchase intent, and a market that rewards depth and precision: which is what a serious content programme produces anyway. It is also the hardest linguistically, which is an argument for doing it while you have budget and attention rather than last.
2. Nordics
Sweden, Denmark, Norway, Finland. Lower localisation cost because of English tolerance, high digital maturity, fast decision cycles relative to DACH, and comparatively thin local competition in many B2B niches. Fast wins that fund the harder markets.
3. Benelux
Netherlands, Belgium, Luxembourg. Netherlands first. Belgium is a two-language market (Dutch and French, plus German-speaking community) and needs a considered decision about whether you serve nl-BE and fr-BE separately or lean on nl-NL and fr-FR. Getting this wrong produces duplicate-intent cannibalisation.
4. Southern Europe
France, Spain, Italy. Higher linguistic investment, more relationship-driven sales, and link ecosystems that are harder for a foreign brand to enter. France in particular rewards genuine local presence.
5. CEE
Poland, Czechia, Romania and others. Growing, less contested, and, in Czechia's case, a market where Seznam still holds meaningful share and needs separate treatment. Covered in the European search piece.
Per-Market Link Ecosystems
This is the part that gets budgeted last and determines outcomes most.
There is no European link graph
A German-language page competing in Germany needs links from German-language domains: trade publications, industry associations (Verbände), regional business press, university and Fachhochschule pages, and specialist community sites. Links from your English domain's existing profile transfer some domain-level authority and very little topical relevance in-language.
What earns links in each region
- DACH: original data, technical depth, standards commentary. German trade press links to substance. It does not link to listicles.
- Nordics: collaborative content, industry benchmark studies, partnerships with local associations. Small, well-connected professional communities.
- France: local media relations, sector-specific publications, and genuine French-language thought leadership. Cold outreach in English performs poorly.
- Spain and Italy: event participation, association membership, and relationship-built editorial. Slower, stickier.
- Netherlands, pragmatic; a genuinely useful tool or dataset earns links quickly.
The operational implication
Budget link acquisition per market, not per programme. A single European link budget always gets spent on the easiest market and leaves the valuable ones unsupported.
Technical Layer: What Still Matters
Since I've argued hreflang is not the problem, here is the short version of what to get right and stop thinking about.
URL structure
ccTLDs give the strongest geo-signal and the weakest authority consolidation. Subdirectories (/de/, /fr/) consolidate authority and are the right default for most brands. Subdomains sit awkwardly between. Pick subdirectories unless you have a specific reason.
Language vs region
de targets German speakers anywhere. de-DE, de-AT, de-CH target regions. Use region codes only when you have genuinely differentiated content: different pricing, different legal terms, different case studies. Region-splitting identical content creates a cannibalisation problem you then have to solve with canonicals.
x-default
Set it. It is your fallback for unmatched language-region combinations and it is routinely missing.
Return tags
Every hreflang annotation must be reciprocated. Non-reciprocal annotations are ignored. This is the single most common technical failure and it is trivially detectable in any crawler.
Don't auto-redirect on IP
Serve a suggestion banner, not a forced redirect. Forced IP redirects break crawling and infuriate the substantial European population that travels or uses a VPN.
The AEO Layer: Answer Engines and European Languages
Generative answer surfaces behave differently by language, and this is currently under-exploited.
English-language queries pull from an enormous, well-covered corpus and the citation competition is brutal. A well-researched German, Dutch or Swedish page on a specific professional topic faces a far thinner corpus, which means a materially higher chance of being the cited source.
Three practical consequences:
- Native-language depth is now an acquisition channel. A genuinely thorough German page on a narrow topic can become the answer engines' default source for that topic in German.
- Structure for extraction. Clear question-formatted headings, direct answers in the first sentence under each, and definitional clarity: in the target language, not translated from an English original that was structured for English readers.
- Entity consistency across languages. Your brand, product names and key concepts should resolve to the same entity across language versions. Inconsistent naming across
/de/and/fr/fragments the entity and weakens citation.
Localisation as a Trust Signal
European B2B buyers read localisation as a proxy for commitment. A German-language site with an English support email, US-format dates, dollar pricing and American case studies communicates that the market is an experiment. This is a conversion issue before it is a ranking issue, and it feeds back into rankings through engagement.
The minimum credible localisation set: language, currency, date and number formats, local contact route, at least one in-region proof point, and legal pages that reflect local requirements. Below that threshold, more content doesn't help.
What To Measure
- Per-market share of voice, not aggregate European traffic. Aggregate hides that one market carries everything.
- Native-language query coverage, the proportion of your target market's high-intent queries where you rank in the target language rather than in English.
- Per-market referring domains in-language. The leading indicator for everything else.
- Citation presence in answer engines per language. Sample manually; it's early and the tooling is immature.
Frequently Asked Questions
Should I use ccTLDs, subdomains or subdirectories?
Subdirectories for most brands, they consolidate authority and are simplest to maintain. ccTLDs make sense where local trust is decisive and you have the resources to build authority separately per domain.
Can I just translate my English content?
You can, and it will underperform. Translation preserves the argument but not the query mapping, the register, or the local proof points. Research the keywords natively, then write to them: using the English original as a source, not a template.
Is machine translation acceptable for SEO?
Machine translation with competent human editing by a native speaker is viable and is what most efficient programmes actually do. Raw machine translation caps your rankings and, more importantly, your credibility with the audience.
How many European languages should I start with?
One, done properly, beats four done thinly. German is usually the highest-value first non-English market for B2B.
Do I need hreflang if my pages are in different languages on different URLs?
Yes, if the pages are equivalents of each other. Hreflang tells search engines which version to serve to which audience and prevents the versions competing.
Why is my German page ranking in the UK?
Usually missing or non-reciprocal hreflang, or an x-default that isn't set. Check return tags first.
How long does a new European market take to show results?
For a competitive B2B market with proper native content and in-market link acquisition, six to twelve months to meaningful organic pipeline. Faster in the Nordics, slower in France and Germany.
Should Belgium get its own pages?
If Belgium is a real revenue market, yes: nl-BE and fr-BE with genuinely differentiated content. If it's incidental, point Belgian traffic at nl-NL and fr-FR and don't create thin duplicates.
Does Switzerland need separate pages from Germany?
If you sell there meaningfully, yes. Different currency, different regulatory environment, different vocabulary, and a distinctly different competitive SERP.
How does GDPR affect multilingual SEO?
Indirectly but materially. Consent rates differ by market, which distorts your analytics comparison between markets. Cookie and privacy pages must reflect national requirements. See the cookie consent piece for why "one EU banner" is a simplification.
Further Reading
If you're planning a European rollout and want the keyword, sequencing and authority work planned together rather than handed to a translation vendor, that's the kind of programme I build. More at younusfardeen.com.