How can I keep only one variation live?

NOTE: Do not set a variation to 0% traffic split. If you do not need it and have more than two variations, Pause it from the right-side menu. If you have two variations and only need one variation live, set both to 50% and then end the experiment. More detailed instructions for this use case can be found below.

If you have a page or experience that has multiple active variations that were created before the launch of the Experiment workflow, they remain live on the URL of the page, but they are not visible in the builder anymore. This is a state of transition that will occur once for a specific page, only until it is transitioned into Experiment.

You cannot start an experiment without choosing at least two variations that will be live as part of it, so if you only need one variation live, follow the steps below.

Go to the Optimize tab and click on Experiment, then on Create Experiment.

Screenshot on 2023-10-26 at 13-17-45.png

Select the page in question, then enter a name for the experiment and a hypothesis. You can type anything here if you are just using this experiment to change which variation is live, without actually running a test for a period of time.

Screen_Shot_on_2020-02-17_at_18_31_43.png

After the experiment is created, set an equal traffic split for all of the variations. It is not important what split you choose as long as you have at least two variations live, including the one you want to keep.

Start the experiment with two or more variations live, then end it immediately. Upon ending the experiment, you will be asked to choose a winning variation. Pick the one you want to keep, and it will be the only variation that remains live on the URL of the page.

Screen Capture on 2023-10-26 at 13-25-59.gif

The workflow above is also valid for any experiment; ending an experiment is the only way to keep a single variation live on the URL of the page. You cannot start an experiment without at least two variations, but you can keep a single one live by ending an experiment.