One brand. A different AI answer in every market.
Cross-border AI search
AI assistants build their answers from local sources in the local language. The authority you built at home does not travel with you. We measure and engineer AI search visibility per market, in five languages.
What is cross-border AI search?
Cross-border AI search is the work of getting a brand understood, trusted and selected by AI assistants in every market it sells in, not only at home. ChatGPT, Gemini and Google AI Mode answer a German buyer from German sources and a British buyer from British ones. A brand therefore has as many AI reputations as it has markets.
It is also called international GEO or multilingual GEO. We treat it as a measurement question first: the home market is the benchmark, and every other market is measured against it.
The same brand scores differently per country
We sample the same brand in several countries, with questions written natively for each market and without the brand name in the question. Five examples from our own panels. Rank is the position among the competitors tracked in that panel.
| Brand | Home market | Abroad |
|---|---|---|
| A Dutch B2B wholesaler | Named in 35% of answers, rank 1 | Germany 23%, rank 2. United Kingdom 18%, rank 4 |
| A Dutch e-commerce software company | 61%, rank 1 | Germany 50% and United States 38%, rank 1 in both |
| A consumer brand in five European countries | 79%, rank 1 | 44% to 69% in the four other markets, rank 1 to 3 |
| A home and living retailer | 14% | Germany 7% |
| An industrial manufacturer | German-language market 15%, rank 1 | United Kingdom 9%, rank 1, on the same 30 questions |
Part of that gap is market position: a brand that is smaller abroad is named less often. The rest is authority, and that part can be engineered.
Why home-market authority does not travel
AI assistants cite local sources. In the five-country panel we listed the 30 most-cited domains per market. Not one domain from the brand's home country appears in any of the four other markets.
| Market | Of the 30 most-cited domains, on the local country domain |
|---|---|
| Home market | 27 |
| Germany | 29 |
| Two Nordic markets | 24 and 28 |
| United Kingdom | 18. The other 12 are English-language .com, .org, .net and .info sites |
The same holds for the brand's own website. The Dutch wholesaler's site is the most-cited source at home, third in Germany and twenty-first in the United Kingdom. In Germany a national supplier directory takes first place.
Every market also has its own trusted institutions. In the Dutch panel of the home and living retailer, the consumer association and the market regulator are among the most-cited sources. In the German panel for the same retailer their place is taken by the consumer advice centres and three German test institutes. In ChatGPT the single most-cited source in that German panel is a test institute for service quality. The press coverage, reviews and test results you earned at home are not what the assistant reads abroad.
Known is not the same as recommended
The gap is not a lack of awareness. When the question names the home and living retailer, the brand appears in 82% of answers in the Netherlands and in 81% in Germany. On category questions without the brand name it is named in 14% of answers at home and in 7% in Germany. The engines know the brand in both markets. What differs is how often they bring it up unprompted.
How local is your category?
Not every category is equally local. We counted how many of the 25 domains ChatGPT cites most sit on the local country domain:
| Category and market | Of the 25 most-cited domains, on the local country domain |
|---|---|
| Home and living retail, Germany | 23 |
| Home and living retail, Netherlands | 20 |
| Industrial B2B, Germany | 16 |
| Industrial B2B, United Kingdom | 10 |
| E-commerce software, Germany | 3 |
| Enterprise software, France | 1 |
In retail and in regulated industry the answer is built from national sources: test institutes, consumer bodies, standards organisations and regulators. In B2B software the same global vendors, analysts and documentation sites are cited in French and German as in English. An enterprise software company in our panels is named about as often in France (3%) as in the United Kingdom (5%) and the United States (4%), and ranks fourth in all three.
That changes the work. A retailer entering Germany needs German proof. A software company needs a place in the global source set first and pages in the local language second: in the French panel a competitor's French-language subdomain is among the cited sources.
The gap is not a law
One online retailer in our panels runs a local domain and a local-language shop in each country. In a first sample it is named at least as often in Germany and France as in its Dutch home market, and its local domain is among the two most-cited sources in all three. Local presence that machines can read closes the gap.
A Google ranking does not close it. One German-language webshop climbed to position 1 in Google on its main terms while its visibility in ChatGPT on market questions fell from 28% to 5% over the same six months.
How we work per market
The method follows four steps from cross-border marketing, translated to AI search. The steps come from the ShoppingTomorrow expert group on cross-border marketing (2025), to which our founder Arjan ter Huurne contributed.
| Step | In cross-border marketing | In AI search |
|---|---|---|
| 1. Market research | Understand the local customer and the local competition | A prompt set per market, written natively. Its own competitor set and its own source landscape |
| 2. Brand positioning | Keep the brand DNA, adapt tone and proof | How each engine describes the brand per market, set against the home-market description |
| 3. Marketing mix | Local channels, creative and targeting | The source mix per market: which media, directories, review platforms and test institutes to earn a place in |
| 4. Digital shop window | A local site, not a translate button | Local pages machines can read: language, hreflang, local facts and local proof |
We never translate a prompt set. Buyers phrase questions differently per language, and competitors, sources and even the framing of a category differ with them.
The Cross-border GEO baseline
For brands in several countries we start with the Cross-border GEO baseline: the home market and two growth markets, measured the same way, for a one-off fee of €1,000. It shows:
- visibility and share of voice per market, on questions without the brand name
- how the engines describe the brand per market, and where that differs from home
- the most-cited sources per market, and the ones where competitors appear and you do not
- a prioritised plan per market across technical, content and authority
From there the work follows our three services: AI Visibility to keep sampling, GEO to improve, and ChatGPT Ads to buy presence where organic visibility takes time. Authority work per market runs through digital PR and mention building in the local language.
Markets and languages
Our team works in Dutch, English, German, French and Spanish. Our first client is headquartered in Paris and sells worldwide. Another is a Swiss manufacturer that operates worldwide. We have sampled AI answers in the Netherlands, Germany, the United Kingdom, France, Italy, the Nordics, the United States and Brazil, and our ChatGPT Ads campaigns run in the Netherlands, Germany, Austria and France.
Three markets have their own page, written in the local language with the sources AI assistants cite there: Deutschland, Österreich und die Schweiz, the United Kingdom and France. For other languages we bring in native specialists from partner agencies in the Everywhere Network.
How we measured this
The figures on this page come from our own panels, sampled up to November 2026: ChatGPT, Google AI Mode, Google AI Overviews, Gemini, Claude, Copilot and Perplexity, depending on the panel, in sessions without personalisation and on questions without the brand name unless stated. Panels differ in size, from about a hundred answers per market to several thousand. The smaller ones show direction, not a precise level. Brands are anonymised.
What is cross-border AI search?
Getting a brand understood, trusted and selected by AI assistants in every market it sells in. Assistants answer per market from local sources, so visibility has to be measured and built per market. It is also called international GEO or multilingual GEO.
Why is my brand less visible in AI answers abroad than at home?
For two reasons. Market position: a smaller brand is named less often. And authority: the press, reviews and directories that confirm your brand at home are rarely cited in another country. In our five-country panel no home-market domain appears among the 30 most-cited sources in any other market.
Can we translate our prompt set and content for other countries?
Translation is not enough. Buyers phrase questions differently per language, the competitors differ and the cited sources differ. We build a native prompt set per market and have content reviewed by native speakers.
Which languages and markets do you cover?
Dutch, English, German, French and Spanish in our own team. For other languages we work with native specialists from partner agencies in the Everywhere Network.
Is every category equally local?
No. In retail and regulated industry, 16 to 23 of the 25 domains ChatGPT cites most sit on the local country domain in our German panels. In B2B software it is 1 to 3 of 25, because the same global vendors and analysts are cited in every language.
Do we need a separate website per country?
Not always. What matters is that each market has local pages a machine can read: local language, local facts and proof, hreflang, and sources in that country that confirm them. In one of our panels a retailer with local domains is named at least as often abroad as at home.
Where do you start?
With the Cross-border GEO baseline: the home market and two growth markets, measured the same way. It shows the gap per market and which local sources to work on first.
What does the Cross-border GEO baseline cost?
A one-off fee of €1,000. It covers the home market and two growth markets: visibility and share of voice, how the engines describe the brand, the most-cited sources and a prioritised plan per market.
See how AI describes your brand in each market
The Cross-border GEO baseline samples your home market and two growth markets, and shows which local sources carry your category. One-off fee: €1,000.