/Docs

Experiment Statuses

Every experiment in A vs B has a status that tells you where it is in its lifecycle. Understanding what each status means — and what you can and cannot do in each state — helps you manage your experiments effectively.

The builder header surfaces the actions that apply to the current status: a Draft shows Publish and Schedule (both gated by the pre-launch checklist), a Running experiment shows Pause, and a Paused experiment shows Resume and Stop. Scheduled and Completed experiments show no launch buttons. All of these require the Publish to production permission.

Draft

A new experiment starts in Draft status. A draft experiment is:

  • Not running — no visitors are being bucketed.
  • Editable — you can change any part of the experiment freely.
  • Not visible to visitors — the snippet ignores draft experiments entirely.

An experiment stays in Draft until you explicitly publish it. Use the Draft state to build and refine your experiment before going live. There is no time limit — an experiment can stay in Draft indefinitely.

Transitions from Draft:Draft → Running (when published) or Draft → Scheduled (when a future launch time is set).

Scheduled

A Scheduled experiment has been configured to launch automatically at a future moment. No visitors are bucketed yet — the experiment is effectively still a draft until the scheduled time arrives.

  • The experiment is hidden from visitors — the snippet ignores scheduled experiments the same way it ignores drafts.
  • The schedule banner on the experiment page shows the launch moment in your local time, the scheduler's timezone, and UTC.
  • You can edit or cancel the schedule from the banner at any point before it fires.

Once the platform's scheduler runs at the assigned moment, the experiment transitions to Running and the scheduler is notified. See Scheduling Experiments for full details and the catch-up window behaviour.

Transitions from Scheduled:Scheduled → Running (when the scheduled time arrives), Scheduled → Draft (when the schedule is cancelled).

Running

A Running experiment is live. Every eligible visitor who lands on a targeted page is evaluated, bucketed, and shown the appropriate variation. Events are tracked and results accumulate in the dashboard.

  • Visitors are being bucketed and shown variations.
  • Results are updating in real time (with a few minutes of processing delay).
  • Some edits are possible but should be done carefully — see Managing Running Experiments.

Transitions from Running:Running → Paused, Running → Completed.

Paused

A Paused experiment has been temporarily halted. No new visitors are bucketed. Existing data is preserved. The experiment can be resumed at any time.

  • Visitors see the original page (the snippet skips paused experiments).
  • All data collected before pausing is preserved.
  • The experiment can be freely edited while paused (since it is not live, there is no risk of corrupting running data).

Use Paused to temporarily halt an experiment if you discover a bug in your variation code, if you are making a major site change, or if you need time to analyze results before deciding whether to continue.

Transitions from Paused:Paused → Running (resumed), Paused → Completed (stopped).

Completed

A Completedexperiment has been stopped and will not run again. No visitors are bucketed. The experiment's full results history is preserved permanently.

  • Variation code is no longer injected for any visitors.
  • All data is preserved and viewable on the results page.
  • The experiment cannot be restarted.
  • The experiment can be archived to hide it from the active list.

Stop an experiment when you have made a final decision: implemented the winning variation permanently in your codebase, or decided the experiment was not successful.

Transitions from Completed:Completed → Archived.

Completing is permanent
Moving an experiment to Completed is irreversible. You cannot restart a completed experiment. If you think you might want to run the experiment again, use Paused instead.

Archived

An Archived experiment is a completed experiment that has been hidden from the main experiments list to reduce clutter. It is non-destructive — all data is preserved.

  • Hidden from the default experiments list view.
  • Accessible via the “Archived” filter in the experiments list.
  • All data is preserved and viewable.
  • Can be unarchived at any time.

Transitions from Archived:Archived → Completed (unarchived).

Status summary

StatusVisitors bucketed?Data collected?Editable?Reversible?
DraftNoNoYes, freelyN/A
ScheduledNoNoSchedule only (edit / cancel)Cancel schedule
RunningYesYesYes, with carePause or stop
PausedNoNo (preserved)Yes, freelyResume
CompletedNoNo (preserved)NoNo
ArchivedNoNo (preserved)NoUnarchive