Skip to content
SilktideHelp

Choosing location and language

Every region in Presence pairs a country with a language. The country is where Silktide asks AI assistants and search engines your questions from. The language is what it asks them in. You choose both when you set up Presence, and again whenever you add a region.

This page helps you make that choice. In brief, choose your languages carefully, and treat the country as a smaller decision. Research so far finds that the language of a question changes which brands AI assistants recommend far more than the country it is asked from. The exception is a question about a place, such as "best dentist near me", where location matters a lot.

Every measurement is made from somewhere

There is no option to measure from nowhere. To find out what ChatGPT, Gemini or Google says, Silktide has to ask them, and each question has to come from somewhere. Choosing a country does not limit who your results are about. It decides where the questions are asked from.

So a university in the Netherlands that recruits international students does not need to measure from every country those students live in. It can measure from the Netherlands, in English, and see what an English-speaking student is told.

Most of the platforms Presence measures take the region's country into account. A few do not accept a location at all, and on those only the language counts.

If you check an answer yourself, ask in the same language and, ideally, from the same country. Assistants such as ChatGPT estimate your location from your connection, so the same question asked from another country can come back a little different.

Language changes the answer a lot

Ask an assistant "which university should I study at?" in French and again in German, and you can get two different lists. That is reasonable. The language you write in suggests where you live, what you can study in, and which institutions you might have heard of. Assistants pick up on that the way a person would.

Studies in 2026 found this again and again. Changing only the language of a question changed which brands were recommended, often by more than asking the same question twice. The effect is strongest for brands that are well known in one country: they are named in almost every answer in their home language, and rarely in English. Global brands are named at similar rates in either language.

So a brand can do well in one language and badly in another. That is one of the most useful things Presence can show you, and it is why each language you care about needs its own region.

Location matters less, unless the question is about a place

For questions that do not depend on a place, such as "what is the best project management software?", the country makes much less difference. When researchers asked the same questions in the same language from different countries, the brands recommended changed far less than when they changed the language. Some assistants barely changed at all. Others, including ChatGPT and Google's, cited more sources from the country they were asked from, while still naming many of the same brands.

Questions about a place are different. Ask for "a good credit union" or "a dentist near me" and assistants answer with businesses in the area they think you are in. If your customers ask local questions, name the place in your and prompts, the way your customers would ("dentists in Austin").

Assistants are also becoming more local over time. Search engines took the same path: early on, a search for "pizza" returned much the same page everywhere, and today Google shows different results on opposite sides of the same city. We expect location to matter more each year, and we will update this page as it does.

What the research says

This is a young field. The studies below are small and recent, they do not always agree, and the assistants they measured change every few months.

  • Language changes which brands are named. Asking about 66 European brands in twelve languages, moving from English to a brand's home language raised how often a locally known brand was recommended far more than it did for a global one (The Language Blind Spot, 2026). Asking coffee questions in four countries, answers in the local language and in English shared fewer brands than two runs of the same question did (Same question, different language, different brands, 2026). With ChatGPT, the language of the question decided whether local suppliers appeared at all, while the country decided which country's suppliers were named (The Language of the Question Selects the Market, 2026).
  • Language moved brands more than country did. Asking the same questions in French, German and English from France, Germany and Switzerland, asking in a brand's own language raised its share of mentions by 12 to 15 points, against 3 to 5 points for asking from its own country. Perplexity cited almost the same sources from any country, while ChatGPT's sources changed with the country about as much as with the language (Cross-border AI visibility study, 2026). Asking 60 questions from India, the UK and the US, Perplexity cited identical sources in all three, while ChatGPT and Google AI Mode cited more sites from the asking country (Asked from India, answered from America, 2026).
  • Location matters for local questions, and on some assistants more than others. Across 31,200 answers, ChatGPT's recommendations for "near me" questions changed almost completely from city to city, and barely changed for questions that did not depend on place. Gemini changed its sources by country even for questions that did not mention a place (Location, Location, Generation, 2026).
  • The assistant can matter more than either. One audit across six European languages found much bigger differences between assistants than between languages, and showed how easily a small test exaggerates a language gap (Cited or Invisible?, 2026).
  • The assistants say they use location. OpenAI says ChatGPT estimates your general location from your IP address to find local results (OpenAI Help). Google uses location, language and device to tailor Search results (Google Search Help).

Search engines care more about location

The search half of Presence is different. Search engines already tailor results to where the searcher is (Google Search Help), so location matters more there than it does for assistants.

Even so, measuring every place and language you care about would cost far more than it would tell you. For most organisations, one region in the right language covers most of what matters. Adding a second country in the same language usually adds little, and can double the number of you measure.

Why more regions cost more

Every region asks its own copy of your questions, so each one you add multiplies how many prompts you measure, and prompts are what your plan pays for. Unlike a web search, every question put to an AI assistant has a real cost. That is why setup asks about language first and treats the country as a smaller choice.

When measuring more countries makes sense

Measuring the same language from several countries is worth it when:

  • Your customers ask local questions, and you compete in more than one place.
  • You are a large brand in several countries that share a language, such as the United Kingdom and the United States, and small differences between them matter to you.
  • You want to be ready for the day assistants become more local, and the extra cost is small next to what is at stake.

For most organisations, one country per language is enough for now.

In practice

  1. List the languages your audience asks in. Each one is a region.
  2. Pick one country for each language. Usually where you are, or where most of that audience is.
  3. Add another country in the same language only when you have a reason, such as local questions or a strong presence in that country.

You can add regions later under Settings → Regions, if your plan allows.

Last updated

Was this page helpful?