Survey variants & A/B
Split traffic across question wordings or designs and compare completion rates.
The Variants tab splits traffic across alternative versions of the same survey, so you can find out which wording or design people actually finish.
What a variant changes
A variant owns its own questions and design. Targeting, display rules and integrations are shared by the whole survey — you are testing the ask, not the audience.
While a variant is selected, the Build and Design tabs edit that variant only. Switch back to Base to edit the original.
Traffic split
Weights are relative, not percentages: 1 and 3 means 25% / 75%, and so does 10 and 30. Each visitor is assigned once and stays with the same variant for the rest of their session, so nobody sees the survey change under them.
With only one variant, that variant takes all the traffic and the base is never shown — which means there is nothing to compare. Add at least two.
Reading the result
The responses view breaks completion rate down per variant and tells you whether a difference is real or noise. Responses are always attributed to the variant that was shown.
Removing a variant
Removing a variant deletes its questions and design. Responses already collected under it are kept and stay attributed to it, so your historical numbers do not move.
Testing the surrounding page rather than the survey itself? That is an A/B test, not a survey variant.