Download websites that use Intersection Observer

About Intersection Observer

This entry needs its meaning explained before its number is any use. Intersection Observer is a browser API rather than a product. Standardised from 2016 onwards and available as Open Source technology with nothing to buy or install, it lets JavaScript code be told when an element enters or leaves the viewport, which is the mechanism behind lazy loaded images, infinite scroll, advertising viewability measurement, and animations that begin when you scroll to them.

Because the interface is built into browsers, a site using it downloads nothing and announces nothing. What we can detect is the polyfill, the standalone script that supplied the same interface to browsers that lacked it, along with the wrapper libraries that bundle that script. The count below is therefore not a count of sites using the capability, which would be enormous and largely unmeasurable. The count is a count of sites still shipping a compatibility shim for it.

As of October 2026 we detect Intersection Observer on 10,242 sites, 5,472 of them resolving to United States and the domains concentrated under .shop. Understood correctly, a polyfill population is a conservatism index. Shipping one means a build toolchain that still targets old browsers, or a template that has not been revised since the shim was necessary, or a dependency quietly carrying it along. All three describe a site whose front end is older than its content.

Version data supports that reading. We record 10 distinct versions, and the release we encounter most often is 0.10.0, dating from the period when the shim was genuinely required. A polyfill nobody needs is also a polyfill nobody has a reason to update, so versions in the field freeze at whatever was current when the project was set up. Version spread here measures the age of build pipelines rather than the pace of a project.

For research the useful framing is inverted. Nobody wants a list of sites using a standard browser feature. A list of sites still carrying its polyfill is a different and more valuable object: a sample of front-end codebases that have not been rebuilt, useful to anyone estimating technical debt across a sector, sizing the market for modernisation work, or studying how long compatibility code outlives the incompatibility that justified writing it.

The limits are the same facts that make the data interesting. Sites using the native interface are invisible to us and always will be, so this is a deliberately partial view rather than a failed attempt at a complete one. Polyfills compiled into an application bundle under a generic filename are frequently missed as well. Anyone treating the figure as adoption of the underlying capability will be badly wrong, and anyone treating it as evidence of an ageing front end will usually be right.