Download websites that use libphonenumber

About libphonenumber

Nothing about a telephone number is as simple as it looks. Length, prefix, whether a range is mobile or fixed line, and whether a leading zero survives when the number crosses a border are all decisions a regulator made and can unmake. libphonenumber exists because of that, and because telephone numbers are one of the few common data types where naive validation is guaranteed to be wrong. Published by Google under Apache 2.0 and originally written in Java in 2009, the library parses, validates and formats numbers according to the numbering plan of every country that has one.

The important property is that the library is mostly metadata. Its parsing logic is small and stable, and the bulk of any release is a table describing which prefixes exist, which lengths are legal, which ranges are mobile and which country a number belongs to. Numbering plans change constantly. Regulators open new mobile ranges, retire old prefixes and lengthen subscriber numbers, so a copy of the library ages in a specific and measurable way. An outdated build does not crash. An outdated build quietly rejects real customers whose numbers were issued after that copy was published, a failure that never appears in an error log and shows up only as conversion the business cannot explain.

Version data makes the ageing visible, which is why this entry is worth more than a name check. We can separate 60 distinct releases in the field, with 1.12.41 recurring across a large part of the population. The spread is effectively a map of how stale phone validation has become across the web, because the release someone bundled is the release their signup form is still running, and bundles get rebuilt only when something else forces a rebuild.

Reading the population correctly requires one large caveat. The library is a dependency rather than a service, and it runs on the server in most of the places where it matters, inside signup pipelines, customer data platforms and messaging systems no crawler can see. What we detect is the browser side slice, the port shipped into a page so a form can validate before submitting. As of October 2026 that slice covers 29,100 sites, and the real installed base across all languages and runtimes is considerably larger and not observable from outside.

Geography reads the same way and deserves the same care. 25,613 of the detected sites are in United States, with the domain mix leaning to .com. Concentration like that says more about which market builds international phone entry into its forms, and where those sites are hosted, than about where the library is genuinely relied upon, because cross border validation is a concern for companies selling across borders and companies like that are not evenly spread.

A per domain lookup answers whether one form validates numbers properly. The list answers questions that only exist in aggregate. Which parts of the web run stale numbering metadata and would reject a newly issued mobile number, how quickly a given industry rebuilds its front end bundles, which sites do enough international business to bother with the problem at all, and which platform vendors freeze a single release into the product they resell to thousands of merchants. For a communications provider, a fraud team, a form vendor or anyone measuring dependency freshness at scale, the version spread across the whole population is the finding, and no single site can produce it.