Ask a generic AI chatbot "Will my premium skincare brand sell in Vietnam?" and you get a generic guess. AI Market Twin runs simulations on a persona pool that's actually grounded in real government and public statistics — Vietnam's Bộ Y Tế food registration process, Indonesia's BPS household income distribution, Japan's 厚生労働省 cosmetics regulations, and 21 more.
Income distribution, profession-level wages, household composition, consumption patterns, and core regulations — measured data pulled from each country's government and public statistics offices is the ground truth for persona generation.
You'll get an answer either way. The question is whether the answer is something you can actually take into a boardroom.
There are two ways to stop an AI from making things up that sound right: ground it in official data, and score it against what actually happened.
We went back to the moment 42 brands were deciding where to expand, analyzed each using only what was known then, and compared the result with the market where the brand actually succeeded. No cherry-picking: all 42 were scored.
Confidence levels. Every result carries a High, Moderate or Low confidence level. High requires several AI models to name the same market with a clear margin over the runner-up; when they disagree or the supporting data is thin, the level goes down. High means check this market first, Moderate means keep it on the shortlist, Low means it needs more verification.
The figures come from a single measurement on September 17, 2026 (42 cases); the method and its limits are in the methodology PDF. Every result PDF lists the data sources used and how the score was built, and engine improvements reach existing customers at no extra cost.
From input to PDF report in about 6–30 minutes depending on the tier. Stages are cross-run several times, split across different AI models.
Checks the product details and decides which jobs, incomes and ages the virtual consumers should have, following each country's statistics.
Inspects sales bans and labelling rules per country and category. Markets where launch is structurally impossible are auto-excluded.
Matching consumers are reused from the accumulated pool and only the gap is newly generated. Each gives a purchase intent and a first-person comment.
Aggregates intent distribution, rejection factors, and trust signals across all virtual consumers to score each market on demand, CAC, and competitive intensity.
Pricing is run five times separately and the median is used, so one run can't swing the result. Price sensitivity is broken out per market.
Any uploaded packaging or ad images are analyzed too. A final check confirms the recommended market and its evidence agree before the result is finalized.
Consensus analysis of a K-food product as a PDF (6 cross-runs · 1,200 virtual consumers · 3 candidate markets · 27 pages) — no sign-up, download directly.