There are two things you can check. A product check answers four questions about one item — where it's made, where it ships from, who the retailer really is, and where your money ultimately goes. A brand check asks a different question: who ultimately owns the company behind the label. Both cite their sources, they are metered separately, and checking a brand never spends a product check.
Product checks run in the browser extension while you shop, or in the mobile app when you're in a store. Brand checks run in all three — extension, app, and here on the site.
Check a product
Browser extension · Chrome, Edge, Brave
Install once. On Amazon, Target, Walmart and 30 other major retailers, a four-line badge shows origin, shipping, retailer and beneficial ownership — with a confidence percentage and every source cited.
Your first 20 checks are free. No account required.
Check a brand
An American-sounding brand is not an American company. A brand check asks who ultimately owns it — separately from any product — and keeps what the public record says apart from what we conclude.
On the website a brand check needs an account. In the extension it does not.
Check a product anywhere
Mobile app · iOS and Android
In a store looking at a shampoo bottle. Standing in a friend's kitchen. Wondering about the URL someone just texted you. The app answers in seconds — five different ways in.
Public launch imminent. Get emailed the moment it ships.
Where it works
Same score, same accuracy, same one-click badge on every one. Grouped by what you'd buy there.
Don't see the retailer you shop at most? Tell us which one to add next.
What every score tells you
Two are about the product. Two are about the money. Every score covers all four, with a country flag and a confidence percentage on each.
Where the physical product was manufactured. From labels, disclosures, customs records, and brand-family knowledge.
Where the actual inventory sits before you buy it. Often a domestic fulfillment center regardless of manufacture.
The company that takes your payment. First-party retailer, or a third-party marketplace seller with its own country of registration.
The parent company that ultimately receives the profit. Brand names roll up to holding groups; we surface the top of the chain.
How it works
We answer four questions about a product by pulling signals from many independent places — not just the retailer's own page. Claude AI (Anthropic's frontier model) reads all of it, weighs it, and produces the four scores with confidence bands. The AI never invents evidence — every citation on the score card links to a real source. The more sources agree, the more confident we are. When they don't agree, we say so.
See a worked example of a score card, with every element labeled — including what the colors mean.
No single source decides a score. On every product we can, we consult:
The assembled evidence is sent to Anthropic's Claude AI, which weighs it against its own trained knowledge of brands and companies — it knows Bosch is German, Hydro Flask is American, TCL is Chinese. Before Claude looks at the structured signals, we require it to write down what it already knows about the brand. That protects against paper trails: a Chinese seller registering "BRAND INC" in Delaware would otherwise fool a naive trademark lookup into reporting "United States." When world knowledge and paperwork disagree, world knowledge wins.
The strongest evidence always wins. A literal "Country of Origin" field on the product page beats a guess from the brand name. An independent registry record beats a hunch. And a company's headquarters can never, by itself, be used to claim where a product is manufactured — a common and misleading conflation we refuse to make.
Each indicator ships with two numbers: a percent (our certainty for the top country) and a confidence (our overall confidence in the whole estimate). These are not the same, and the distinction is the point.
For all four indicators, we score Hong Kong, Macau, and Taiwan as their own countries — never folded into mainland China. This is a hard rule.
Every score has a Contest this finding button. Tell us what's wrong and (if you have one) attach a source — a manufacturer's page, an official disclosure, a photo. Our review team verifies every claim. When we agree, the correction becomes the new answer for every future shopper, marked ✓ Verified. The dataset gets better because you tell us when it isn't.
Our scores are well-supported probabilistic estimates, not factual claims. When evidence is strong, we are confident. When evidence is thin, we say so loudly. Every source is cited, and every citation is a link you can click.
Want to go deeper? The full technical specification of every data source lives at DATA_SOURCES.md in our public repository.
Your first 20 checks are free. Plans from $4.99/mo.