The two-line convention that quietly connects a million subscriptions

Every time someone types a domain into a feed reader and clicks subscribe, the reader performs a small, invisible act of archaeology. It fetches the homepage, parses the HTML, and looks for a <link> element carrying rel="alternate" and a type of application/rss+xml or application/atom+xml. If it finds one, the user never has to know a feed URL exists. That is feed autodiscovery — a convention so mundane it rarely gets named, and so load-bearing that removing it quietly breaks thousands of workflows.

The mechanism was proposed in 2002 and works by placing a handful of attributes in a page's <head>. The reader retrieves the page, scans for the marker, resolves the href to an absolute URL, and opens the feed. No directory lookup, no manual path-guessing. Feedly, Inoreader, NewsBlur, and every serious feed reader built in the last decade implement it as a basic expectation — not a feature.

Terminal window on a dark monitor showing a raw Atom feed in XML, angle brackets and namespaces visible, person's hands on keyboard
A feed is a text file sitting on a server. Between the publisher and the reader there is a fetch, and nothing else.Photo: Tima Miroshnichenko / Pexels
Printed copy of RFC 4287 on a wooden desk, a hand resting on the first page, coffee mug beside it
Specifications are argued over in rooms like this one. Atom's took a working group two years; RSS 2.0's had a single author and a closing date.Photo: Kindel Media / Pexels

Where it breaks

The fragility is almost always introduced at the CMS level. A theme update strips the <head> partial. A caching layer serves a stripped HTML skeleton. A JavaScript-rendered site delivers the link element only after client-side hydration — too late for a reader making a straightforward HTTP GET. WordPress, which powers a large share of the web, emits the tag correctly by default, but plugins that override wp_head can silently remove it. Ghost publishes the element reliably. When a newsletter platform migrates a publication to a custom domain, the autodiscovery tag is often the last thing tested and the first thing missing.

The consequences are quiet rather than catastrophic. A site's feed still exists; autodiscovery's absence just means the reader cannot find it unassisted. Subscribers who already know the feed URL are unaffected. New subscribers who paste the homepage URL into a reader get an error, assume there is no feed, and leave. The site loses readers it never knew it was losing.

From the record

How the tag looks

  • The raw HTML element<link rel="alternate" type="application/rss+xml" title="Feed" href="/feed.xml"> — the marker a reader looks for in the page's head

Where it fails

  • Autodiscovery usually fails in one of three ways: JS-rendered pages that add the tag only after the browser runs scripts, theme updates that strip wp_headoutput, and platform migrations to custom domainsthat drop the tag.

What publishers can verify

Validation is trivial using the W3C Feed Validation Service, which checks both feed syntax and whether the source page carries a correct autodiscovery tag. A page can carry multiple <link> elements pointing to different formats — RSS and Atom simultaneously — and readers handle the redundancy gracefully, typically preferring whichever appears first. Two lines of markup, placed once, permanently extend a site's discoverability to every compliant reader on the open web.