Download websites that use Google Font API

About Google Font API

Google Font API is a design decision that turned into a compliance question. A stylesheet link pulls typefaces from a vendor's network, the visitor's browser makes a request to that network to fetch them, and the visitor's address travels with the request. European courts and data protection regulators have treated that transfer as something a site owner has to justify, and the practical result is that a font embed is now a legal fact about a website as well as a typographic one.

The service arrived in 2010 and is operated by Google on Freemium terms, which in practice means the typefaces are free to use and the hosting is free to consume while the vendor keeps the traffic. As of September 2026 we detect the embed on 36,998,303 sites, filed under Font scripts because the mechanism is what we can see rather than the intent behind it. What we match is a reference to the vendor's font endpoints in the markup or the stylesheets of the page we fetched.

The population that matters is often the complement. A privacy consultancy, a law firm building a case list, a regulator sampling a sector, or a company selling a self hosting alternative all want the domains that still pull fonts from a remote network, narrowed to a jurisdiction where the practice is contested. Answering that for one site takes a browser and a network tab and is useless as the basis for a programme of work. The download answers it for a country code suffix, an industry list or a customer portfolio in a single pass.

United States accounts for 17,306,518 of the sites carrying the embed and the suffix weight sits on .com, which is a reminder that our country column follows serving infrastructure. A German or Austrian business hosted in a data centre elsewhere will not be counted under the market whose rules apply to it, and that is exactly the case a jurisdictional analysis cares about. Segment on the domain suffix, the language of the page and the address published on the site before drawing any legal conclusion from the country column alone.

There is a second and far less adversarial use for the same field. Font delivery is one of the few visible traces of how a site was assembled, so the embed is a decent proxy for template driven construction, and its presence beside a page builder or a theme framework tells you the site was put together rather than designed. Template marketplaces, agencies and performance tooling vendors segment on that signal, because a team that self hosts its fonts and a team that hot links them are usually different teams with different budgets.

The undercount here is real and worth stating. Self hosted copies of the same typefaces are invisible to us, because the files are ordinary assets on the site's own domain with nothing marking them as having come from the vendor's library. Proxying through a delivery network, bundling at build time or inlining the stylesheet hide the relationship just as effectively. So the field is reliable when it says the remote service is being used and unreliable as a measure of who uses these typefaces at all. As a record of visitor data leaving for a third party, the column is exactly what it claims to be. As a typography census, it is not, and nothing in the file pretends otherwise.