Experiment Lifecycle
Experiment Lifecycle
Every experiment in Elevate moves through a series of statuses as it progresses from creation to completion. This page explains each status, what happens in it, and what actions are available.
Status Flow
Draft → Preview → Running → Done
↓ ↑
Paused ────┘An experiment can also be deleted from any status.
Statuses
Draft
The experiment has been saved but isn't fully configured yet. You can come back to it later to finish setup.
What's happening:
The experiment is saved in your dashboard but not active
No traffic is being split, no visitors are affected
You can edit any aspect of the experiment freely
Available actions:
Edit the experiment
Submit for review (moves to Preview)
Delete
Preview
The experiment is fully configured and ready to be reviewed before launching. Some experiment types require additional setup steps in this status.
What's happening:
The experiment is complete but not yet live
No traffic is being split
For price tests (duplicate products): this is where you complete the review checklist — copy reviews, sync integrations, and submit for Elevate review
For other test types: you can QA and verify everything looks correct
Available actions:
Launch the experiment (moves to Running)
Mark as Complete (if you decide not to launch)
Sync experiment data
Duplicate
Delete
Running
The experiment is live and actively collecting data. Visitors are being split between variations.
What's happening:
Traffic is being split according to your allocation settings
The web pixel is tracking page views, add to carts, checkouts, and purchases
Statistical significance is being calculated as data comes in
Results are updating in real time on the Results and Advanced Analytics tabs
Available actions:
Pause the experiment (moves to Paused)
Mark as Complete (moves to Done) — available unless the experiment has already reached statistical significance
Sync experiment data — refreshes metafields and analytics
Duplicate
Delete
Note: Once an experiment reaches statistical significance, the "Mark as Complete" option is replaced with guidance to either apply the winner or end the test. This prevents accidentally ending a successful experiment without acting on the results.
Paused
The experiment has been temporarily stopped. Traffic splitting is halted, and all visitors see the original experience.
What's happening:
No new visitors are entering the experiment
Existing data is preserved — nothing is lost
The experiment can be resumed at any time
Available actions:
Resume the experiment (moves back to Running)
Mark as Complete (moves to Done)
Sync experiment data
Duplicate
Delete
When to pause:
You noticed an issue with the variation and need to investigate
You're running a promotion or sale that could skew results
You need to make changes to your store that would affect the experiment
You want to temporarily stop the test without losing data
Tip: Pausing and resuming is fine for short periods. Long pauses (weeks) can affect data quality since visitor behavior may change over time.
Done
The experiment is complete. Traffic splitting has stopped, and all visitors see the original experience.
What's happening:
The experiment is no longer active
All data and reports remain fully accessible
For price tests with duplicate products: duplicates are automatically archived in Shopify
You can review results, export data, and reference the experiment at any time
Available actions:
Resume the experiment (moves back to Running) — if you want to collect more data
Apply Winner — implement the winning variation's changes (available for applicable test types)
Duplicate — create a new experiment based on this one's configuration
Delete
Deleted
The experiment has been permanently removed.
What happens:
The experiment is removed from your dashboard
Test data is cleaned up
Metafields are updated to remove the experiment configuration
For price tests with duplicates: duplicate products are archived
How Pausing Affects Data
When you pause an experiment, Elevate stops the data collection window. The time your experiment was paused is excluded when calculating:
Experiment duration — Paused time doesn't count toward the displayed duration
Projected revenue — Visitor velocity calculations account for paused periods
Statistical significance — Only data from running periods is included
This means pausing doesn't corrupt your data — it just extends the calendar time needed to reach significance.
Resuming from Done
You can resume an experiment even after marking it as complete. This is useful if:
You ended the test too early and want to collect more data
Results were inconclusive and you want to let it run longer
You want to re-run the same experiment after making store changes
When resumed, the experiment goes back to Running status and continues from where it left off — existing variation assignments are preserved.
Where You See Status
The experiment status is displayed:
On the experiment card in your dashboard (with a color-coded badge)
At the top of the experiment detail page
In the Experiment History timeline (Raw Data tab), which logs every status change with timestamps
Last updated