Ask AI in Norwegian, Get Norwegian Brands: Language Now Picks the Market

By Daniel Okafor

The language of a question decided which brands existed in the answer: asked in Norwegian, ChatGPT named three Norwegian accounting tools and no American ones. Asked in English minutes later, from the same connection in Oslo, it did the reverse.

The Norwegian query returned Fiken, Tripletex and Conta. The English version of the identical question returned FreshBooks, QuickBooks and Wave. The Norwegian tools had vanished.

The buyer had not changed. Same city, same product category, same machine. The language alone selected the market.

What did the study actually test?

A researcher in Tallinn ran 234 controlled tests across four countries and six languages, posting the results to arXiv. The counts came back lopsided. Where the question used the country's official language, a global brand appeared in just one of 24 runs. Asked in English from Istanbul and Tallinn, a local supplier appeared in none of 12.

The model was not short of knowledge. In two of six English answers on the Estonian connection, it noted the user was in Estonia and that local tax rules argued for a local system — then named five foreign products anyway. It knew. It did not say.

Earlier work by the same researcher quantified the effect. Across 12,933 answers, the language of the question explained 26.5% of the variation in responses. Brand identity explained 1.5%.

The study has stated limits: one model, six runs per test, and an author who sells AI visibility measurement for a living and discloses that in the paper. The author's own framing — read it as a loud signal rather than settled law.

Why does English behave like its own market?

These systems assemble answers from what they have read, sorted by language. An English question pulls from the English web, where global brands have spent two decades accumulating coverage. A Norwegian question pulls from Norwegian pages, where local tools get reviewed and compared.

English does not even act as a neutral default. On the Turkish connection, English queries returned zero Turkish suppliers, while every other language tested returned local results.

A separate paper comparing Chinese and Western models named the pattern: the existence gap. A brand documented in one language exists for the models that read that language and barely registers for the rest.

What does this do to visibility reports?

For a US company selling only to English speakers, nothing. For everyone else, it changes what the monthly report counts. The study's author put it plainly: a company selling into Germany or Norway that checks its AI visibility in English is checking a market it does not compete in, and its real rivals are invisible to its own instrument.

The reverse hits local champions. They own the local-language answer and disappear the moment a buyer types in English — which is how procurement teams and international buyers tend to type.

What does fixing it cost?

The author, who runs ESBO Ltd, a link building and brand mentions agency with a publisher network in 53 languages, reports pricing data across 16,625 publishers:

• Median sponsored article in Spanish: $655 across 1,642 sites
• Polish: $480
• Central and Eastern Europe: about $490
• American-audience English sites: $452
• British English sites: $1,040
• Swedish, Norwegian, Finnish and German: $1,230 or more

The Nordic languages and German cost the most, yet far fewer brands compete for those placements than for English ones, according to the agency's experience.

Translating your own website helps less than expected. What you publish about yourself is the weakest signal AI has — the Estonian suppliers in the study already publish English documentation, and it did not move the English answer. What moves it is third parties naming you in the buyer's language: local press, roundups, review platforms and podcasts whose transcripts the models read.

Measurement needs the same discipline. Take the five questions a buyer would ask about your category, translate them — machine translation for a first pass, a native speaker for the second — and run each several times per language. The study found the top recommendation changed across identical runs on four of six prompts, so one screenshot is one draw from a distribution. Log which brands appear in each language and which sources get cited.

The person in Oslo got a competent answer in both languages. The difference was which companies were allowed to exist, decided by a choice nobody made deliberately. Buyers make that choice daily, and a company that only checks in English has no idea what they see.