Download websites that use FrontPage

About FrontPage

FrontPage is more interesting to us dead than it ever was alive. The editor first shipped in 1995, was aimed at people who did not want to write HTML by hand, and Microsoft discontinued it long before most of the pages we still find were last touched.

A page produced by the editor today is almost never a page anyone maintains. The tool left recognisable markup, proprietary server extensions and characteristic generator strings behind, and finding those on a live domain usually means the file has simply not been replaced since it was saved. For anyone studying the ageing edge of the web, the technology works as a dating instrument rather than as a product.

As of October 2026 we still record FrontPage on 43,849 sites. The generator strings on those pages resolve into 21 distinct versions, with 4.0 recurring across a large part of the set, which is itself informative. A living product shows a version distribution that moves with every release cycle. A discontinued one shows a fossil layer, frozen at whichever release the author happened to own on the day the page was written, and the shape of that layer is the clearest evidence we have that these sites are static rather than merely conservative.

Geography follows survival rather than adoption. Germany accounts for 14,091 of what we see, which surprises people who expect an American product to leave a mainly American footprint. Small business and association pages built in one market and never migrated stay online at very different rates depending on how the local hosting industry handled its ageing shared plans, and the country breakdown here measures that far more than it measures who once bought the software. Domains ending in .com carry the bulk of the list.

Caveats matter more for a dead technology than for a live one. Generator tags can be copied, templates were passed around freely, and a modern site can inherit a fragment of old markup without any trace of the editor remaining underneath. Microsoft kept the product Proprietary and there is no repository to reconcile our fingerprints against. The count is best understood as evidence of unreplaced pages rather than of installed software.

Two kinds of buyer want the file. Security and risk teams want the domains still serving output from software that stopped receiving patches, because unmaintained pages tend to sit on unmaintained hosting, and the exposure is rarely limited to the page itself. Agencies, hosts and registrars want the same domains as rebuild and migration candidates, since an owner who has not touched a site in that long is either gone or overdue. Either way the unit of work is the whole set, filtered by country, suffix and neighbouring technology, which is a query the dataset answers and a per-domain lookup cannot.