This article was AI-generated as part of an experimental historical-content project. The date reflects the period being analyzed rather than the date the article was originally written.
When Google announced last week that AI Overviews would roll out in more than 100 countries and territories, the headline number was users: more than 1 billion a month. The more interesting sentence was about language.
“If you’re in any country with AI Overviews,” Google wrote, “you can now get them in any of the currently supported languages, including English, Hindi, Indonesian, Japanese, Portuguese, and Spanish.” Its example: a Spanish speaker in the United States can now see AI Overviews in Spanish.
That small change separates two things that used to travel together, the country a person searches from and the language they search in. For reputation, that separation matters more than the country count.
Four variables, not one
It helps to break a search into its parts. When someone looks up a company, at least four things are in play:
- Location: where the searcher is, which affects which version of Google they get.
- Interface language: the language Google is set to show them.
- Query language: the language they actually type in.
- Source language: the language of the pages the answer is built from.
In classic search, these mostly lined up. A user in Mexico, searching in Spanish, got a page of mostly Spanish-language results, many from Mexican outlets. Reputation teams organized their work the same way, by market.
With August’s expansion to six countries, AI Overviews arrived with local language support in each one. October’s change goes further. A user can now be in one country and get an overview in a different supported language. The geography of a first impression is no longer the geography of the searcher.
What Google has not said
Google has not explained how an overview in Spanish, shown to someone in Texas, chooses its sources. It might lean on Spanish-language pages, English-language pages, or a mix, and the answer probably varies by query. I would not assume any particular behavior.
What can be said is simpler. A generative answer can only be as good as the material it draws on, and the material about most companies is uneven across languages. A U.S. company might have deep English coverage, a short Spanish-language Wikipedia article last updated two years ago, and a handful of Spanish-language news stories about one product recall. If a Spanish-language overview draws mainly on that second set, the user gets a fluent summary of a thin and possibly skewed record.
Reputation geography
This is why I think of it as reputation geography. The relevant map is not where your offices are. It is the set of languages your customers, employees, investors and critics actually search in, wherever they live.
For a U.S. consumer brand, that almost certainly includes Spanish. For a company with a large workforce in India, Hindi may matter even for searches made in North America. For an investor audience, the picture may look different again.
The practical work follows from that map, and it differs from the market-by-market checks many teams began this summer. Start with audiences, not countries: list the languages your key stakeholders search in, including the ones spoken inside your home market. Run the same questions in each language from the same location, since that is now a realistic scenario. Then compare the overviews side by side. Where one language version of your story is noticeably thinner or older, that is the gap a generative answer will fill with whatever it can find.
Google says the rollout began last week. Many companies have spent this year asking what AI Overviews say about them. The better question now is what they say in each language your stakeholders use, and whether anyone on your team can read the answer.