Do I need a separate project per country?

If the competitor set differs, yes. The same brand in two markets usually meets different rivals, and that is the test rather than the flag on the project.

Does setting a city limit the prompts to local questions?

It gives the local context those prompts need. A local prompt without a place is answered for nowhere in particular.

What does the vertical actually do?

It tunes how answers are parsed and which prompt patterns are suggested. Pick the nearest fit; Other is honest and costs you only the suggestions.

Market, language and vertical

The context fields on a project: market country, language, city or metro, service and vertical, and what each one changes about your measurements.

Updated October 20, 2026

Five fields describe the market a project is being measured in. None of them change a stored run after the fact, and all of them change how new runs are read and what gets suggested.

Market country and language

Models answer the same question differently depending on where and in what language they think the question comes from. A UK buyer asking about insurance gets a different set of brands than a US one, and that is not noise.

Set both to the market you sell into. If you sell into several and the competitor set differs between them, use a project per market: the competitor list is the thing that cannot be shared, and the whole comparison rests on it.

City or metro, and service

These are for local and service businesses, where the buying question carries a place.

"Best emergency plumber" is unanswerable. "Best emergency plumber in Bristol" names three companies. The city and service fields are what let the prompt set be built around questions a local buyer would actually ask, and they are what the suggested prompts use.

A national SaaS brand leaves both blank.

Vertical

A long list, from local trades through SaaS and B2B services to medical, legal and manufacturing. It tunes two things: how answers are parsed for brand names, and which prompt patterns are suggested during setup.

Pick the nearest fit. The cost of Other is losing the suggestions, not losing accuracy.

Industry benchmarks

With a vertical set, the project can be read against industry benchmarks: how presence in your category looks in general. Useful for the first conversation about whether a score is good, and never a target. Your competitor set is the comparison that matters, because that is the set you actually lose deals to.

Tracking cadence

The project-wide default for how often prompts run, which individual prompts then follow. It is the setting that scales your model spend, so it lives next to the projected cost by engine rather than off on its own. See run cadence.

Note

These fields are the same ones an agent can read and write over MCP as the project's context, so a setup conversation can fill them in and a later session can read them back instead of asking you again. See brand and voice context.

Questions about market, language and vertical

Do I need a separate project per country?: If the competitor set differs, yes. The same brand in two markets usually meets different rivals, and that is the test rather than the flag on the project.
Does setting a city limit the prompts to local questions?: It gives the local context those prompts need. A local prompt without a place is answered for nowhere in particular.
What does the vertical actually do?: It tunes how answers are parsed and which prompt patterns are suggested. Pick the nearest fit; Other is honest and costs you only the suggestions.

Was this page helpful?