Activate & Manage Experiments
Previewing an experiment
Previews require the tracking code to be installed and working - without it, previews won't function.
The main preview - from the Reporting page
Once an experiment is created, its Reporting page shows the primary way to preview it. Under Traffic allocation, each variation (Control, Variation #1, etc.) has its own Preview link.
Previews work through a URL query parameter that includes the experiment ID and variation ID. Since it's just a URL parameter, you can append it to any page on your website to preview that variation there too - not just the page the experiment is set up on.
Previews inside the experiment editor
While editing, the Variations tab also has its own preview - useful for checking your changes as you make them, before saving the experiment. This works the same way across A/B tests, Personalization, Overlays, and Surveys, though Surveys require saving a draft first before the Preview button appears.
Previewing unpublished or unsaved changes
You can preview an experiment even before it's published, or after making changes you haven't saved yet.
-
Open the experiment you want to preview.
-
If it's unpublished, or has unsaved changes, a Preview option appears - click it for a real-time preview.
-
Review the preview and make any adjustments before finalizing and publishing.
Activating an experiment
From the Experiments section of your dashboard, press the Play button on the right side of any experiment to make it live.
You can also Pause, Reset, Delete, adjust Settings, Download data, or jump to Create Report directly from the experiment's own Reporting page.
Resetting an experiment
Resetting clears an experiment's results so you can start measuring from scratch. Press Reset either from the experiments list or from inside the specific experiment, then confirm in the pop-up that appears.
We recommend resetting an experiment whenever you make meaningful changes to it - this way, you can clearly track the impact of your new changes on conversion rate, rather than mixing old and new data together.
Testing an experiment live before launch
Two ways to safely check how an experiment behaves on your live site before real visitors are affected.
Option 1 - Testing segment with 100% traffic
-
In the Audience tab, select Segment → New segment.
-
Find your IP address, then set the condition to Include → IP address → Is → [your IP].
-
In Traffic Allocation & Goals, set your traffic to 100% so you're guaranteed to be included.
-
If you only want to check how the Variation looks, set Control to 0% and Variation to 100%.
-
Publish the experiment.
After any change, allow 4-5 minutes for it to go live. If it still doesn't apply after that, clear your cookies.
Don't forget to change the traffic settings back once you're done testing, or your real visitors won't see the experiment.
Option 2 - Variation Picker in the browser extension
If you'd rather leave your traffic split as-is (e.g. 50/50) instead of setting it to 100%, use the Variation Picker in the browser extension instead. It lets you manually jump between Control and any Variation directly, without needing to change your experiment's traffic settings at all.
This is also a convenient way to demo a variation to a client or stakeholder before launch.
Experiment statuses
-
Active - the experiment is published and currently running.
-
Paused - the experiment is paused and not currently running.
-
Draft - the experiment has been created and saved, but never published.
-
Expired - the experiment ran and reached its scheduled end date.
-
Publishing - Shopify Price Testing only. The experiment is in the process of being published.
Additional statuses - Scheduled, Deployed, and possibly Archived - are planned for a future release and will be added here once available.