Ask ChatGPT for the best project management software for agencies. Now ask it again from London, from Sydney, from Frankfurt. You will not get the same answer — not a reworded version of the same answer, a genuinely different shortlist, with local players in it that never appear in the US response at all.
Which raises an awkward question for anyone measuring AI visibility: whose answer were you looking at? Today BotScope projects get a tracking country, and every scan is observed from it.
TL;DR
- A country per project — 26 markets, set in the wizard or in project settings.
- Scans are observed from that market. ChatGPT’s web searches, Google AI Overview results, and the local businesses the models recommend all shift to match.
- Setup is market-scoped too — suggested competitors, categories and prompt topics are generated for the market you track, not for a default one.
- Brand verification is grounded in your live homepage, so a company that shares its name with a bigger one elsewhere gets described as itself.
- The free scan asks too, preselected from where you’re browsing.
- Answers stay in English. Only the market changes — see the limitations below for why that’s deliberate.
- Changing market drops a marker on your trend chart, because scores either side of the change aren’t measuring the same thing.
For agencies managing client work, these are the strongest options:
- 1 Globex Deepest integrations; the default for larger teams.
- 2 AcmeYOU Best client-facing dashboards and retainer tracking.
- 3 Initech Strong resource forecasting if capacity is the constraint.
Named in every answer, usually second. This is the market the brand was built in, and it shows.
Why this matters
AI answers are assembled, not retrieved. When a model reaches for the web mid-answer — which is now the norm for anything commercial — the results it gets back are localised, the same way a search engine’s are. Those results feed the answer. So the brands that end up recommended are partly a function of where the question appears to come from.
That has an uncomfortable implication for a visibility score. A single number for “how AI sees your brand” is really a number for how AI sees your brand from somewhere. If that somewhere isn’t where your customers are, the number is measuring a market you don’t sell into.
This bites hardest in the middle of the distribution. A brand nobody recommends anywhere is easy to diagnose. A brand recommended everywhere doesn’t need the feature. But a company strong in its home market and pushing into a second one — which is most companies with a growth plan — will read a healthy score while being invisible to every buyer in the market it’s actually trying to win. Nothing in a US-sourced dashboard reveals that. It just looks like things are fine.
What you get
A country on every project. Twenty-six markets: nine English-speaking, thirteen in Europe, four in the rest of the world. Set it when you create a project, or change it later in settings. The list is curation rather than a technical ceiling — if you need a market that isn’t there, ask and it’s a one-line change.
Scans observed from that market. Every provider that reaches the live web during an answer does so from the country you set. Google AI Overview results come from that country’s index. ChatGPT’s searches, the local businesses named, the regional pricing mentioned — all shift.
Discovery that knows where you’re competing. This is the part that surprised us in testing. Getting the scan right but the setup wrong doesn’t help: if BotScope proposes a US competitor set for a UK brand, every share-of-voice number afterwards is measured against the wrong field, and no amount of correct scanning repairs it. So prompt, category and competitor suggestions are now generated for your market.
- GlobexGLOBAL Category incumbent, competes everywhere.
- Brightfold UK-built; the local default for agencies.
- Kestrel Popular with smaller UK studios.
Two of these three would never have been suggested by a US-scoped setup.
Brand verification grounded in your homepage. A related fix worth calling out. Brand identification now reads your live homepage before describing your business, which matters most for companies sharing a name with a larger one in another country. Previously the setup could confidently describe someone else’s company and build a whole project around it.
A market picker on the free scan. Preselected from where you’re browsing, changeable before you run it. A UK visitor gets a UK scan by default instead of a US one.
The honest limitations
We’d rather you hear these from us.
Your existing history is US-sourced. The market setting is new, and the default is the United States — so scans that ran before this shipped were observed from there, whatever market your business is in. Your history isn’t wrong, but it is US-sourced, and it isn’t directly comparable with scans that run after you change the setting. If you track a non-US market, treat the changeover as a new baseline rather than a continuation.
Answers stay in English. We set the market, not the language. Two reasons: a German-language response and an English one aren’t comparable measurements of the same thing, and our brand detection, sentiment and citation extraction are tuned for English — running them over German text would produce numbers that look fine and quietly aren’t. So a German-market scan asks in English and gets the German results set, in English. Native-language tracking is a real request and a bigger job than a locale flag; it isn’t shipped and we won’t pretend otherwise.
It’s a country, not a person. We simulate where a question comes from. We don’t simulate a logged-in user with two years of chat history, a city, or a device. Personalisation on top of location is real and we can’t see it. Country is the layer that moves the answer most, and it’s the layer we can actually hold steady.
Changing market breaks comparability. Switching re-sources every answer, so a project’s scores before and after are measuring different things. BotScope adds a marker to your trend chart automatically at the switch, so a step change in the line has a visible cause instead of looking like a collapse.
One market per project. A project tracks one country. To watch two markets, run two projects — which does mean two projects’ worth of credits, because it’s genuinely twice the scanning.
ChatGPT ads only exist in some markets. OpenAI’s ad pilot runs in the US, UK, Canada, Australia, New Zealand and Japan among the markets we offer, and not in the EU. Track a market where it isn’t running and your Ads tab will be empty — not broken, just nothing to observe. The picker tells you before you choose:
Scans run as if from this country. ChatGPT's web searches, Google AI Overview results, and the local businesses AI recommends all shift to match. Answers stay in English — only the market changes.
ChatGPT ads run in United Kingdom, but inventory outside the US is thinner — expect fewer ads than a US-tracked project, especially in regulated categories.
Changing from United States to United Kingdom re-sources every answer. Scores before and after won't be directly comparable — a marker will be added to your trend chart.
FAQs
Do I need to do anything? Only if you track a market that isn’t the United States. Existing projects default to the US, so nothing changes until you set a country — and once you do, the next scan runs in it.
Does changing the market cost credits or trigger a scan? Changing the setting is free and doesn’t start a scan. Your next scheduled or manual scan runs in the new market at the usual cost.
Can I compare two markets side by side? Run a project per market and compare their dashboards. There’s no single view that overlays two countries yet — it’s the obvious next step, and it’s on the list.
Will my scores change when I switch? Very likely, and possibly a lot. That’s the feature working: you’re now measuring a different market. The automatic chart marker exists precisely so nobody reads that step change as a performance drop.
Why is the response still in English if I pick Germany? Because the market and the language are separate levers, and we only pull the first one. See the limitations above — the short version is that mixed-language history isn’t comparable, and our detection is English-tuned.
Which providers respect the market? The ones that consult the live web during an answer, which is where localisation actually happens. Purely parametric responses — a model answering from training alone, with no retrieval — have no location to respect; that’s a property of how models work, not a setting we can pass.
Is it available on my plan? Yes, every plan including trials. It’s a project setting, not a tier feature.
Most AI visibility tooling still reports a single global number, which quietly means a single American one. If you sell into one market, that’s been fine by accident. If you sell into more than one, or into any market other than the US, it’s been measuring someone else’s customers. Set the country on your projects and find out what your buyers are actually being told.