Download websites that use React Router

About React Router

A routing library is an architectural confession. A site shipping React Router has decided that navigation happens in the browser rather than on the server, which changes how it renders, how it handles history, how it behaves on a slow connection and how search engines see it. As of September 2026 we detect the library on 54,066 sites.

The version field is where the list stops being trivia. We record 12 distinct versions in the wild, and the release that keeps recurring is 5.2.0, which belongs to a generation of the library predating the current API rather than to the line the maintainers now document. Front end code gets upgraded when something forces it, and a router sits deep enough in an application that replacing it means touching every screen. The result is a long tail of applications that work, ship and earn revenue on a release line that stopped gaining features a while ago. For anyone selling migration work, framework consulting, testing tools or a competing router, the size and shape of that tail is the whole opportunity, and it only becomes visible in aggregate.

Presence also identifies a class of site with predictable properties. Single page applications tend to render on the client, which makes them heavier for crawlers, harder to index, more dependent on one JavaScript bundle and more likely to sit in front of an API than a template engine. SEO tooling vendors, performance and monitoring products, prerendering services and accessibility auditors all care about exactly that class, and none of them can identify it from a domain name.

Shopify is the maintainer our record names, and the project has been published under MIT terms since 2014, written in JavaScript, so adopting it involves no purchase and no contract. Adoption follows the React ecosystem instead, which is why United States accounts for 41,222 of the detections and the .com suffix runs through the list. Concentration of that kind reflects where React hiring and React agencies are dense rather than any licensing restriction.

Detection carries the usual caveat and one specific one. The library is recognised from bundle signatures and runtime markers, and a build that inlines and minifies everything can hide both the presence and the release, so the real footprint is larger than the attributed count and the version reading is a sample rather than a full inventory. Applications behind a login are underrepresented for the same reason, since a crawler only sees what a visitor without credentials sees.

Checking one domain tells you what one team chose. The downloadable list tells you what a market chose, and the version column tells you roughly when it chose it.