Do Exact-Match Domains Still Work? A Field Study in Entity Resolution from Danish Local Lead Generation

The conventional wisdom is that exact-match domains (EMDs) are dead — that Google's 2012 update killed the cheap trick of buying plumberlondon.com and ranking on the strength of the URL alone. The conventional wisdom is half right. The string-matching boost is gone. But something the keyword tools cannot see is very much alive, and it is more interesting than the old trick ever was: search engines no longer match the letters in your domain, they resolve the entity behind them. This is a field study of that distinction, built on live data from one of the cleanest natural experiments available — Danish local lead generation, where the capital city has three legitimate spellings and the engine treats all three as the same place.

The natural experiment: one city, three spellings

Copenhagen is written København in Danish — with the letter ø, which is not a decorated 'o' but a distinct letter of the alphabet. The moment you put that city into a domain name, which cannot contain ø cleanly, you are forced to choose how to spell it. There are three real-world options, and all three appear live in the Copenhagen search results:

• koebenhavn — the official transliteration. When ø cannot be used (passports, international systems, URLs), Danish law renders it 'oe'. To a Dane this is the correct, institutional spelling.
• kobenhavn — the lazy spelling. What a tourist, expat or anyone on a non-Danish keyboard types when they drop the ø to a plain 'o'. To a native Dane it reads slightly like a foreigner's spelling error.
• kbh — the native abbreviation. What Danes themselves type on a phone when they want speed. Short, unambiguous, no special characters, instantly read as 'Copenhagen' by any local.

What the keyword tools say — and the trap inside it

Pull the three spellings through Semrush and the raw numbers look like they settle the question instantly. They do not. They set the trap.

The naive read is: 'koebenhavn is dead, never use it.' The trap is assuming search volume on the literal string equals the value of the spelling in a domain. It does not. No Dane types koebenhavn because their keyboard has the real ø — they type københavn. The tool shows ~20 not because the spelling is worthless but because it is a transliteration nobody keys in directly. The question is not 'how many people type these exact letters' — it is 'when someone searches the real query, does a domain spelled this way still rank?' For that you cannot ask a keyword tool. You have to look at what actually ranks.

The field test: a domain spelled 'wrong' that ranks anyway

Here is the cleanest piece of evidence in the whole study. saunakobenhavn.dk is a live aggregator of saunas in Copenhagen. Its domain uses the lazy plain-'o' spelling, kobenhavn. Its Search Console data shows which queries actually earn it clicks.

This is the finding the keyword tool actively hides. The tool said koebenhavn had ~20 searches; Search Console said a kobenhavn domain earns clicks on dozens of københavn searches. The volume on the literal string was never the point. The entity behind it was.

Ranking is not conversion: the second axis nobody measures

If entity resolution means all three spellings rank, you might conclude the spelling no longer matters at all. For ranking, nearly true. For getting the phone to ring — the only thing a lead-gen asset is paid for — it matters enormously, and on a completely different axis: trust. A Dane in a plumbing emergency who lands on vvs-kobenhavn.dk reads the plain 'o' as the fingerprint of a foreign-built spam site and hesitates to hand over a phone number. The same Dane on vvs-koebenhavn.dk or vvs-kbh.dk reads a native, legitimate business. Same ranking, very different conversion.

What the live SERP confirms

The Copenhagen results page for plumbing ('VVS København') is the whole theory rendered in production. Ranking organically you find vvs-kbh.dk and kbhvvsteknik.dk (abbreviation), vvskoebenhavn.dk (oe) and vvskobenhavn.dk (plain o) sitting side by side — every spelling resolved, every spelling ranking. You find vvs-norrebro.dk, an EMD scoped to a single neighbourhood, dodging citywide competition. And running paid ads across the top sits 3byggetilbud.dk, Denmark's dominant lead-generation aggregator — proof that the lead itself is valuable enough to pay a premium CPC for, again and again.

Where the money actually is (and why intent beats volume)

The lead-gen lesson rides on top of the entity finding. Once you stop chasing literal string volume, you start reading the two numbers that matter: commercial intent and what an advertiser will pay for the click. A few from the Danish data:

• elektriker charlottenlund — ~1,900 searches/mo, CPC ~$6.67, but keyword difficulty only ~10%. High volume, premium-paying suburb, almost no SEO competition. That combination is rare and worth more than any head term.
• skadeservice københavn (water/fire damage restoration) — only ~110 searches/mo, but CPC ~$15.52. Two or three leads a month, each a job worth thousands and usually insurance-paid. Low volume, extreme value.
• glarmester københavn (emergency glazier) — ~390/mo, CPC ~$3.00. Tiny volume by Anglo standards, but in a country of 5.9M an urgent local query converts to a call at near 80%.

In a small market, a tool showing 30 searches does not mean 'no demand' — it means a few dozen people a month with an urgent problem and a wallet open. Volume is a vanity metric here; resolved-entity relevance plus intent is the real one.

The general lesson for answer engines

Strip away Denmark and the trades and you are left with a principle that governs GEO and AEO directly: modern engines reward the thing you are about, not the string you spelled. A domain is the crudest possible signal of 'what is this page about', and even there the engine has moved from matching letters to resolving entities. Answer engines do the same thing at a far higher resolution — they decide what entity your content authoritatively represents and cite you on that basis, not on keyword overlap. The EMD boost surviving in entity form is the smallest, cleanest demonstration of the rule every answer-engine strategy depends on.

The playbook

1. Pick the entity, not the keyword volume. Choose the city or district you want to own; ignore the literal-string search count on transliterations.
2. Buy the native or institutional spelling as your primary — '...-kbh.dk' for short and trusted, or '...-koebenhavn.dk' for long-form and email-safe.
3. Buy the lazy spelling defensively and 301-redirect it onto the primary, so you catch foreign/lazy traffic without showing locals an 'error' spelling.
4. Go hyper-local where the head term is contested — each neighbourhood is its own entity and a far cheaper one to rank for.
5. Score niches by intent and CPC, not volume — in a small market the high-CPC, low-volume emergency niche is the asset, not the head term.
6. Then earn the entity. The domain is a starting nudge; the ranking and the citations come from genuinely being the authoritative answer for that place and service.