Experimentation
URL redirect tests
Test a whole page against a whole page, rather than one element against another.
A redirect test sends part of the eligible audience to a different address instead of editing the page in place. It is the right shape when the variant is a genuinely different page: a rebuilt landing page, a new template, a campaign destination that already exists and only needs measuring.
The mechanics of assignment, weighting and reporting are identical to a multi-variant test. Traffic is split by variant weight, the assignment is sticky, and both arms report into the same statistics screen against the same primary objective.
Redirects need more care than in-page edits, because moving a visitor to another URL moves them out of, or into, the audience of anything else running. Pertento treats that as a blocking collision: a redirect that overlaps another running experiment is reported before either can be automated, since it breaks the independence both readings rely on.
The redirect is applied by the runtime already on the page, so there is no server rule to add and nothing to unwind when the test ends.