The languages each check reads, and the markets that use them
A store writes its policies, its stock labels and its footer in its own language. Most of the scanner’s checks find their subject by matching words, so each one works only in the languages on its own list — and they are not the same list. Below is every check, the languages it reads, and what that leaves for each market. None of it is typed in. It is computed from the same tables the scanner runs, so it cannot drift from them.
The totals
- Languages every check reads
- 36
- Read by some checks only
- 0
- Not read at all
- 1
- Markets covered
- 69
Of 37 checked. Read by all 3 checks that work from a word list, before any AI layer is involved.
The narrowest: Finding the policy page and Reading the policy page, at 36 of the 37. A scan says so on the report instead of passing the store on a check that never looked.
Thai. A store in Thailand is left to the AI layers for every check here.
Merchant Center markets with at least one language every word-list check reads.
Check by check
3 checks work from a word list, so each is limited to the languages it was given. When a page is in another language the report says the check could not look, and the store is neither passed nor failed on it.
| Check | What it looks for | Languages it reads |
|---|---|---|
| Finding the policy page | Recognising a privacy-policy link label or web address written in the language, so the scanner reaches the page at all. | 36 of the 37 below |
| Reading the policy page | Recognising the data-protection wording that makes a reached page a real policy rather than a placeholder. | 36 of the 37 below |
| A period of time on a policy page | Spotting a return window, a dispatch time or a delivery estimate written as a number and a day word. | 37 of the 37 below |
| The facts inside a policy | Reading a stated fact — a return window, a delivery time, a shipping cost — out of the page in whatever language it is written in, and quoting the sentence back. | Any language |
The last row is the exception, and it is not counted in the totals above. It reads 16 stated facts — on the shipping, returns, privacy, contact and terms pages — in whatever language the page is written in, and has to quote the sentence it read them from. It is left out of the count because it runs on an AI provider that fails open: when the provider is down it returns nothing, and nothing looks the same as looked-and-found-none.
The two policy layers, kept apart
Finding a policy page and reading it are different jobs, done by different rules, and they fail separately. One tick would hide which half failed, so the table keeps two columns.
Finds the policy page
Can the scanner recognise a link label or a web address for a privacy policy written in this language? If it cannot, it never reaches the page.
Reads the policy page
Having reached it, can the scanner recognise the data-protection wording that makes it a real policy rather than a placeholder? If it cannot, it reports a page it has just found as empty.
Handled: the rule works in this languageNot handled: the row is marked
That gap is not hypothetical. For one morning the scanner could find a Bulgarian privacy policy by its phrase for personal data, and then report the page it had just found as containing no data-protection wording. This table exists to keep that kind of gap visible.
Language by language
37 languages, A to Z. Each row shows the two policy layers, then how many of the 3 word-list checks read the language at all.
| Language | Finds the policy page | Reads the policy page | Checks that read it | Markets that use it |
|---|---|---|---|---|
| Arabic | Finds | Reads | 3 of 3 | Saudi Arabia, United Arab Emirates, Egypt, Bahrain, Jordan, Kuwait, Lebanon, Oman |
| Bulgarian | Finds | Reads | 3 of 3 | Bulgaria |
| Catalan | Finds | Reads | 3 of 3 | Spain |
| Chinese | Finds | Reads | 3 of 3 | Taiwan, Hong Kong |
| Croatian | Finds | Reads | 3 of 3 | Croatia |
| Czech | Finds | Reads | 3 of 3 | Czechia |
| Danish | Finds | Reads | 3 of 3 | Denmark |
| Dutch | Finds | Reads | 3 of 3 | Netherlands, Belgium |
| English | Finds | Reads | 3 of 3 | United States, United Kingdom, Ireland, Canada, Australia, New Zealand, Singapore, South Africa, India, Nigeria, Kenya, Ghana, Philippines |
| Estonian | Finds | Reads | 3 of 3 | Estonia |
| Finnish | Finds | Reads | 3 of 3 | Finland |
| French | Finds | Reads | 3 of 3 | France, Belgium, Canada, Switzerland |
| German | Finds | Reads | 3 of 3 | Germany, Austria, Switzerland |
| Greek | Finds | Reads | 3 of 3 | Greece |
| Hebrew | Finds | Reads | 3 of 3 | Israel |
| Hungarian | Finds | Reads | 3 of 3 | Hungary |
| Indonesian | Finds | Reads | 3 of 3 | Indonesia |
| Italian | Finds | Reads | 3 of 3 | Italy, Switzerland |
| Japanese | Finds | Reads | 3 of 3 | Japan |
| Korean | Finds | Reads | 3 of 3 | South Korea |
| Latvian | Finds | Reads | 3 of 3 | Latvia |
| Lithuanian | Finds | Reads | 3 of 3 | Lithuania |
| Malay | Finds | Reads | 3 of 3 | Malaysia |
| Norwegian | Finds | Reads | 3 of 3 | Norway |
| Polish | Finds | Reads | 3 of 3 | Poland |
| Portuguese | Finds | Reads | 3 of 3 | Portugal, Brazil |
| Romanian | Finds | Reads | 3 of 3 | Romania |
| Russian | Finds | Reads | 3 of 3 | Kazakhstan, Uzbekistan, Belarus |
| Serbian | Finds | Reads | 3 of 3 | Serbia |
| Slovak | Finds | Reads | 3 of 3 | Slovakia |
| Slovenian | Finds | Reads | 3 of 3 | Slovenia |
| Spanish | Finds | Reads | 3 of 3 | Spain, Mexico, Argentina, Chile, Colombia, Peru, Ecuador, Uruguay, Paraguay |
| Swedish | Finds | Reads | 3 of 3 | Sweden |
| Thai | Not found | Not read | 1 of 3 | Thailand |
| Turkish | Finds | Reads | 3 of 3 | Turkey |
| Ukrainian | Finds | Reads | 3 of 3 | Ukraine |
| Vietnamese | Finds | Reads | 3 of 3 | Vietnam |
Scope, precisely: this measures which languages each word-list check can look in, not how well it judges what it finds. A check that cannot read a page publishes a row saying so, and that row is left out of the store’s score — our gap costs the merchant nothing, it just stops paying them for it.
Markets covered
69 Merchant Center markets have at least one language that every word-list check reads. Take the two policy layers on their own and that number is 69: a store there can expect its privacy policy to be found and read, while the checks in the table above still say they could not look.
- Argentina
- Australia
- Austria
- Bahrain
- Belarus
- Belgium
- Brazil
- Bulgaria
- Canada
- Chile
- Colombia
- Croatia
- Czechia
- Denmark
- Ecuador
- Egypt
- Estonia
- Finland
- France
- Germany
- Ghana
- Greece
- Hong Kong
- Hungary
- India
- Indonesia
- Ireland
- Israel
- Italy
- Japan
- Jordan
- Kazakhstan
- Kenya
- Kuwait
- Latvia
- Lebanon
- Lithuania
- Malaysia
- Mexico
- Netherlands
- New Zealand
- Nigeria
- Norway
- Oman
- Paraguay
- Peru
- Philippines
- Poland
- Portugal
- Romania
- Saudi Arabia
- Serbia
- Singapore
- Slovakia
- Slovenia
- South Africa
- South Korea
- Spain
- Sweden
- Switzerland
- Taiwan
- Turkey
- Ukraine
- United Arab Emirates
- United Kingdom
- United States
- Uruguay
- Uzbekistan
- Vietnam
Russia is not a market here, on purpose. Google suspended advertising services there in 2022, so a store there cannot run Shopping ads. Russian is in the table above because it is the working language of Kazakhstan, Uzbekistan and Belarus, which are active markets.
What this page does not tell you
How accurately the scanner judges what it reads is a different measurement, with its own dated evidence.
See accuracy and method