Each experiment can have one of these five states:
An experiment in draft has never been run and can be edited by all users except users with the “browse” level permissions.
Running experiments are called “Active”; they can contain a mix of active and inactive variations (variations can be stopped manually or automatically). Active experiments cannot be placed back in a Draft state. Any warnings about problems with this test will be available after 24-48 hours (depending on the project).
Nothing will be running on the website when a whole test is paused. It can be resumed at any give time without missing data.
This is a special state. The experiment is completed when the following conditions are true:
There was a variation winning with 97% confidence (or the confidence value you have set) and the minimum run-time (optional) was met
All the variations automatically turn off and the original turned out to be the best variation (by default variations are turned off when they have a negative impact on your site and 97% confidence); this can be changed in the Experience Summary under Settings.
The maximum run-time of the experiment was met without a clear winner.
Note: The winning variation may be kept running (this is an optional setting)
After an experiment is Paused or Completed, you can change the state to Archived. This will pause the winning variation that may have been running during the Completed state. An Archived experiment can never be started again, but it can be Cloned and/or Deleted.
Note: Experiment variations in Draft status can be edited. You cannot add new variations to experiments that have any status other than Draft. Also, depending on your Role, you may not be able to edit current variations once an experiment has moved out of Draft status.