Download websites that use RSS

RSS has been written off many times and keeps turning up in scans anyway. The format dates from 1999, it publishes a site's recent items as a machine readable list, and the reason it survived has little to do with people using feed readers. Publishing platforms emit a feed automatically, and the link element pointing at it stays in the page whether or not a single person ever subscribes. So the marker measures something quite different from what it appears to measure.

As of October 2026 we detect a feed declaration on 23,556,737 sites, an open format carrying Open Source terms, with no vendor and no licensing story to complicate the analysis. The number should not be read as an audience. The number is a count of sites built on software that expects to publish a stream of items, which is a description of what a site is for rather than a description of its technology stack.

The distinction turns the column into a filter with several distinct buyers. Media monitoring companies and aggregators need the addresses of every site that publishes structured items, because a feed is a cheap and permitted way to follow a source at scale. Podcast platforms need the subset whose feeds carry audio enclosures. Teams assembling text corpora need much the same list for entirely different reasons. And anyone selling to publishers can use the marker to separate sites that produce content on a schedule from sites that put up a page once and walked away.

For commercial prospecting, the interesting move is again the crossing rather than the column. A feed beside a store platform marker suggests a shop with a content operation attached, which is a different buyer with a different budget from a shop without one. A feed beside a corporate site with no other publishing markers usually means a neglected news section the marketing team inherited and stopped feeding. Both patterns are visible in a file and invisible in a lookup, because both are comparisons rather than facts about a single domain.

United States accounts for 10,550,707 of the sites declaring a feed and .com carries the suffix weight, a distribution that follows publishing platforms and hosting rather than reading habits. Our country attribution comes from serving infrastructure, so a national media study should be built from the suffix, the language of the page and the site's own contact details, with the country column used as a first cut and nothing more.

The honest caveat is that a declared feed is not a live one. We record the declaration in the page. We do not fetch the feed, count its items, or check when it was last written to, so an abandoned blog and a newsroom publishing several times an hour appear the same way in the file. Feeds that exist at a conventional address without being declared in the markup are missed entirely, which pushes the count in the other direction. Treat the column as a strong indicator that a site was built to publish and a weak indicator that it currently does. Verifying the second part is a per domain job, and the file is what tells you which domains are worth that job.