Markets and languages
Answer engines answer differently in different places. A Hungarian asking ChatGPT about bank accounts gets Hungarian sources; someone in Austria asking in German gets Austrian ones. So each site has markets: a country and the language people there ask in, or a language with no country for a site that sells everywhere.
| Site | Markets | Asked |
|---|---|---|
| a .hu shop | Hungary · Hungarian | in Hungarian, from Hungary |
| a .com software company | Global · English | in English, from nowhere in particular |
| a shop with de-DE, de-AT and de-CH pages | Germany, Austria and Switzerland · German | in German, from each country |
How markets are chosen
Section titled “How markets are chosen”Unless you set them, Vercatus detects them, and the site’s setup page shows the evidence:
- The domain’s country ending: .hu is Hungary, .de Germany, .co.uk the United Kingdom.
- hreflang alternates: a few regional versions (de-AT, de-CH) become markets; dozens of them mean an international site, asked globally.
- The language of the pages: a language with a real share of the pages becomes a market too.
A generic ending (.com, .io) with one language and no regional signal is global in that language. Detection runs again with every full run, from the pages it read.
Setting them by hand
Section titled “Setting them by hand”On the site’s setup page, under Markets, an admin chooses Set markets by hand and picks up to six: a country (or Global) and a language for each. Use detection returns to the detected ones. A topic rebuild then writes topics for the new markets.
What changes with the markets
Section titled “What changes with the markets”- Topics. Each topic belongs to one market, with phrasings in that market’s language, written the way people there ask (local offers, prices and providers where it is natural). The plan’s topics are shared between the site’s markets, so adding a market does not add topics; it spreads them. See Topics.
- The poll. Each question is asked of every answer engine from its market: the engines are told where the person asking is, and Google’s AI answers are read from that country’s results page. See The poll.
- AI demand. Demand is estimated per topic and market. A market the model has no parameters for is estimated from another market, with a wider range and grade D. See Topics and AI demand.
- The site audit. A run writes its own questions per market, in that market’s language, and shares the run’s questions between the markets.