Split URL testing

Split URL testing is a form of A/B testing where each variant lives at its own distinct URL, and visitors are randomly redirected to one of them, rather than a single page serving different content to different visitors via client-side or server-side logic. It predates most modern experimentation tooling and is still used when variants differ so substantially, a full page redesign or a different technology stack, that maintaining them as one dynamic page isn't practical.

The tradeoff against a standard, single-URL A/B test is real: a redirect adds latency and a possible flash of the wrong page, it complicates analytics because traffic and links now point to multiple URLs that need to be consolidated in reporting, and it can affect search engine indexing if the separate URLs aren't handled with proper canonical tags or redirects. A single-URL test avoids all of this by rendering the chosen variant directly, which is why split URL testing is now the less common approach outside of major redesigns.

It remains the practical choice when a variant cannot be expressed as a content change on an existing template, for example testing an entirely new checkout flow against the current one, and it is still valid for measuring conversion rate and other outcomes as long as redirect-based sample loss and tracking consistency across the two URLs are accounted for.