Download websites that use Ghost

About Ghost

A site built on Ghost is usually a business rather than a page. The software was written for publishers who intend to charge readers, and it ships with membership, newsletters and paid subscriptions in the core rather than as add-ons. Finding it on a domain says something about intent: somebody chose a platform whose central feature is billing an audience, and turned down both the general purpose Blogs route and the closed newsletter services that own the reader relationship.

The choice splits the footprint into two very different populations. One part runs on the managed service sold by Ghost Foundation, where updates arrive on their own and the operator never touches a server. The rest is self-hosted, because the code is released under the MIT license and anyone can run it on their own machine. From outside the two look identical, and version data is the only reliable way to tell them apart.

As of October 2026 we detect Ghost on 11,377 sites and record 317 distinct versions among them, clustered around 6.65. A spread of releases that wide, for a platform that updates itself when it is managed, is a direct measurement of the self-hosted tail: installations pinned to whatever release was current when somebody deployed them and never revisited afterwards. For a security researcher the spread is the whole story, because an old release keeps serving traffic long after the weakness it contains is public.

Geography reinforces the reading. 5,140 of the sites sit in United States, a smaller share than products sold through a national sales force tend to show, and .com carries the bulk of the names. The comparatively even spread fits a platform bought by independent writers, small media outfits and technical teams everywhere rather than by one market, and the software is written in JavaScript, which puts it within reach of the same developers who would otherwise build something custom.

The publishing model also shapes what surrounds the software. A site charging for access needs payment processing, an email delivery path and a membership database, so the platform rarely appears alone and the technologies detected beside it describe the business more precisely than the platform does on its own. An installation with no payment provider anywhere near it is a blog. An installation wired into billing and bulk email is a publication with revenue, and the difference is visible in the data without anyone having to guess.

Detection depends on markers the platform leaves in the page: the generator tag, script and asset paths, and the shape of the theme layer. A publisher who strips those markers, or who serves the site as a static export or through an aggressive proxy, will not be counted, and a heavily customized theme can hide the version even when the platform itself is visible. The count is a floor, and the version breakdown is a floor within a floor.

People who want the full list rather than a single lookup are after the population, not the domain. Hosting companies use it to find self-hosted installations worth migrating. Theme and plugin developers use it to size a market that has no app store to count. Security teams use it to scope who is exposed to a given release. Media researchers use it to map independent publishing outside the platforms that report their own numbers. None of those questions can be answered one domain at a time.