Geographies & languages

AI answers are local. The same question asked in two countries returns different brands, different sources and often a different language. AgentMention asks your prompts in the market you configure.

Where the country is set

The country that actually scopes your answers lives on the prompt group, not on the brand. Every group has one, and it decides the market its prompts are asked in — which is to say, which competitors and which sources the answers draw on.

Your brand also carries a country, but its role is narrower: it is the default applied to new groups, set during onboarding and editable in Settings. Changing it does not move existing groups.

A group's country can be changed from the group itself — unless it is locked. Groups created from a suggested topic inherit that topic's country and keep it: the topic and its prompts were built for that market, so re-pointing them elsewhere would measure a question nobody asked there.

Changing a group's country changes what its future answers look like. Data collected before and after describes two different markets and should not be read as one trend.

Language

Prompts are asked in the language you write them in. Mention matching works across languages, but only for the forms you have configured — which makes name variations the practical lever here.

If your brand is written differently in another script or language, add each form as a variation. A brand transliterated in the answer and not listed in your configuration is a mention that does not count, and its absence looks exactly like a visibility problem. See Brands.

Working with several markets

A group per market is the unit that matters: each market gets its own group, with its own country and its own prompts in the local language. Since every metric is calculated per group, this alone keeps the numbers readable — a single group mixing two markets produces one average that describes neither.

Name the groups so the market is obvious at a glance; you will be switching between them constantly.

A brand per market is the heavier option, and it is worth it when the markets differ in more than their prompts:

• Different competitors. The competitor set belongs to the brand, so two markets with genuinely different rivals need two brands to be measured fairly — Share of Voice is calculated against that set.
• Different brand names. A brand trading under another name or script in a market needs its own name and variations.
• Separate reporting. A brand per market gives each one its own dashboard, audits and trend, with no filtering discipline required.

Where the competitor set and the name are the same, groups are enough. Switching between brands is a single control in the app. See Brands for how multiple brands work and who can access them.

Reading geography-driven differences

• A prompt group where you are absent everywhere but at home is a coverage finding, not a content-quality one. The sources in that market do not describe you.
• Local sources dominate local answers. Expect the Sources leaderboard to look almost unrelated between countries — that is the normal case, and it is why a market needs its own plan.
• Language-specific phrasing matters. Translating a prompt literally often produces a question nobody asks. Write local prompts locally.

Related

• Prompts tab — building a local prompt set
• Brands — when a market needs its own brand
• Sources tab — the local source landscape