Download websites that use Partytown

About Partytown

A site usually adds a technology to gain a feature. Partytown is added to undo a cost the site has already taken on, which is what makes finding it worth something. The library moves third party scripts, the analytics beacons and tag managers and advertising pixels, off the browser main thread and into a web worker, so that marketing code stops competing with the page itself for the resources that decide how fast a visitor thinks the page is.

Presence therefore says two things at once about the operator. The site carries enough third party tagging for the weight to be noticeable, which points to an active marketing organization rather than a brochure page, and somebody on the team measured that weight and went looking for a remedy. Voluntary performance work of this kind is not something a template ships by accident, and it usually sits alongside an in house front end capability. Built by Builder.io and published under MIT, the library is self hosted, so nobody signs up for an account, nobody lands in a vendor customer list, and the only way to count the installed base is to look at pages.

Several kinds of buyer want that population written down. Performance monitoring and real user measurement vendors are selling to teams who have already admitted they have a speed problem. Tag governance and consent platforms are selling to the same teams from a different angle, because a site with enough scripts to need a worker also has enough scripts to need an inventory of them. Edge and content delivery vendors want the sites willing to change their delivery path, and agencies pitching performance retainers want the sites that already tried a fix on their own and may need help finishing it.

As of October 2026 we detect Partytown on 4,786 sites. Detection reads the worker bootstrap and the configuration attributes the library leaves in the document, so a deployment that renames the served path or inlines a modified build can slip past us. The honest reading of the number is a floor rather than a census, and the same caution applies to any library a site is free to rebuild before serving.

For a researcher the gap between a full extract and a per domain lookup is the gap between a yes and a distribution. A lookup confirms one site. The extract, filtered to the Performance group and cross referenced against everything else the same host runs, answers the questions that actually move a decision: which tag managers and analytics vendors sit behind the worker, what frameworks the adopting sites are built on, whether adoption clusters in commerce or in publishing, and how the population shifts between snapshots. Partytown arrived in 2021, recently enough that the deployed base is still changing shape, and a client library written in JavaScript with no account behind it leaves crawl data as the only place that change is visible at all.