Deleting Experiments

Removing experiments from PressPlay is straightforward, but understanding when deletion is appropriate and what happens to your data ensures you maintain a clean, organized experiment history without losing valuable insights.

When to Delete Experiments

Safe to Delete

Consider deleting experiments in these situations:

• Duplicate Experiments: Accidentally created duplicates can be removed without consequence

• Misconfigured Tests: Experiments with fundamental configuration errors that can't be fixed

• Obsolete Hypotheses: Tests based on hypotheses that are no longer relevant due to product changes

• Superseded Experiments: When a better version of the same test has been created

• Test Experiments: Practice experiments created while learning the platform

Think Twice Before Deleting

Avoid deleting experiments in these cases:

• Completed Experiments: Finished experiments contain valuable data. Keep them for historical reference

• In-Progress Tests: Use the stop function instead of delete to preserve partial results

• Failed Experiments: Even failed tests provide learning value. Keep them to avoid repeating mistakes

• Low Priority Tests: Instead of deleting, lower their priority or change status to NOT_READY

Deletion Rules by Status

PressPlay enforces status-based deletion rules to protect data integrity:

Can Delete Directly

These statuses allow immediate deletion:

Cannot Delete Directly

These statuses have special deletion behavior:

Why These Restrictions?

Status-based deletion rules exist to:

• Prevent accidental data loss from running experiments

• Preserve completed experiment data for analysis and learning

• Maintain audit trail of all executed tests

• Avoid race conditions during experiment transitions

• Ensure statistical integrity by preventing deletion of partially-run experiments

Design philosophy: Once an experiment starts running and collects real user data, it should be preserved as a historical record, even if stopped early. This ensures your team can learn from all experiments, including ones that were stopped.

How to Delete an Experiment

Standard Experiment Deletion

1. Navigate to your app's experiment list

2. Locate the experiment you want to delete

3. Verify the experiment is in NOT_READY, READY, or ERROR status

4. Click the experiment to open details

5. Click "Delete Experiment" button (usually in top-right or menu)

6. Confirm deletion in the popup dialog

7. Experiment is immediately removed

Warning: Deletion is permanent. There is no undo or recovery option.

Attempting to Delete In-Progress Experiments

Important: You cannot delete a running experiment. When you attempt to delete an IN_PROGRESS experiment:

1. The system automatically changes its status to STOPPING instead

2. The experiment is not deleted, only stopped

3. Status transitions from STOPPING to FINISHED

4. Once FINISHED, the experiment cannot be deleted

Key point: If you want to delete an experiment, you must do so before it starts running (while in NOT_READY, READY, or ERROR status). Once an experiment enters IN_PROGRESS status, it becomes part of your permanent experiment history.

Screenshot Order Group Deletion

Screenshot order experiments are created in groups with linked internal_sso_id. Deleting these requires special handling:

Cannot Delete Individual Experiments

If an experiment is part of a screenshot order group, attempting to delete it individually will fail with an error message. You must delete the entire group.

Deleting Screenshot Order Groups

1. Navigate to screenshot order experiments

2. Identify the internal_sso_id for the group

3. Use "Delete Screenshot Order Group" function

4. System deletes all experiments in the group simultaneously

Requirements:

• None of the experiments in the group can be IN_PROGRESS

• If any experiment is running, you must stop it first

• Once stopped (FINISHED), the group cannot be deleted

What Happens When You Delete

Immediate Effects

• Experiment disappears from your experiment list

• Backlog queue reorders (if deleted experiment was READY)

• Priority slot becomes available for other experiments

• Next experiment in queue may start if it was blocking queue progression

Data Deletion

When you delete an experiment:

• Experiment record: Permanently removed from database

• Variants: All variant configurations deleted

• Settings: Experiment-specific settings removed (shared settings remain)

• Assets: Uploaded assets (icons, screenshots, etc.) remain in storage but are no longer linked

• Performance data: Any collected performance data is deleted

• Audit logs: Deletion event is logged for security audit purposes

What Is NOT Deleted

• Shared experiment settings used by other experiments

• App-level configuration

• Other experiments (even with same priority)

• Historical data from FINISHED experiments (since these can't be deleted)

Data Retention Policy

Finished Experiments

Completed experiments (FINISHED status) cannot be deleted and are retained indefinitely. This ensures:

• Complete experiment history for analysis

• Ability to review past decisions

• Compliance with data retention requirements

• Long-term learning from test results

Stopped Experiments

When you manually stop an IN_PROGRESS experiment, it transitions to FINISHED and is retained permanently, including:

• Partial results collected before stopping

• Reason for stopping (if provided)

• Complete configuration and settings

Alternatives to Deletion

Change to NOT_READY

Instead of deleting a READY experiment:

• Change status to NOT_READY

• Removes from backlog queue

• Preserves configuration for potential future use

• Can be reactivated by changing back to READY

Lower Priority

Instead of deleting low-priority experiments:

• Assign very low priority (e.g., 1)

• Keeps experiment in system for reference

• Effectively defers indefinitely

• Can be promoted if conditions change

Archive Concept (Future Feature)

While not currently available, many users request an "archive" feature to hide experiments without deletion. Check PressPlay updates for this potential enhancement.

Deletion Best Practices

Delete Promptly

If you identify duplicate or mistaken experiments, delete them immediately. This prevents:

• Confusion when reviewing experiment list

• Accidental execution of wrong experiment

• Cluttered interface

Review Before Deleting

Before confirming deletion:

1. Double-check you're deleting the correct experiment

2. Verify you don't need any configuration details from it

3. Consider if lowering priority or changing to NOT_READY is better

4. Ensure it's not part of a screenshot order group you want to keep

Communicate with Your Team

If multiple people manage experiments:

• Discuss before deleting experiments created by others

• Document why experiments were deleted

• Establish team guidelines for when deletion is appropriate

Keep Completed Experiments

Never try to "clean up" FINISHED experiments. They cannot be deleted, and that's by design. Use filters to hide completed experiments if they clutter your view.

Recovering Deleted Experiments

No Recovery Available

PressPlay does not provide a way to recover deleted experiments. Deletion is permanent.

Prevention Strategies

Since recovery isn't possible:

• Always confirm deletion carefully

• Use NOT_READY status instead of deleting when uncertain

• Take screenshots of configuration before deleting if you might need details later

• Consider creating a "template" experiment at low priority instead of deleting

Recreating Deleted Experiments

If you accidentally delete an experiment:

1. Create a new experiment

2. Manually reconfigure with the same settings

3. Re-upload assets if needed

4. Set appropriate priority

5. Change status to READY when complete

Common Deletion Scenarios

Scenario: Created Experiment with Wrong Asset Type

1. Verify experiment is in NOT_READY, READY, or ERROR status

2. Delete the incorrect experiment

3. Create new experiment with correct asset type

4. Configure variants and settings

5. Set to READY when complete

Scenario: Want to Remove Old Queued Experiments

Instead of deleting:

1. Review each experiment to determine if it's still relevant

2. Delete truly obsolete experiments (in READY status)

3. Change marginally relevant experiments to NOT_READY

4. Keep potentially useful experiments at low priority

Scenario: Need to Clean Up After Testing Platform

1. Filter experiments by date or name to identify test experiments

2. Verify they are in NOT_READY, READY, or ERROR status

3. Delete test experiments one by one

4. Leave FINISHED experiments—they'll serve as learning examples

Scenario: Experiment Started but Was Misconfigured

1. Attempt to delete or explicitly stop the IN_PROGRESS experiment

2. System changes status to STOPPING (deletion is not performed)

3. Wait for status to transition to FINISHED

4. Cannot delete; experiment is now part of permanent history

5. Create new correctly configured experiment

6. Add notes to both explaining what happened

Deletion and Backlog Management

Impact on Queue

Deleting a READY experiment affects the backlog:

• Experiment is removed from priority queue

• Remaining experiments maintain their priorities

• Next experiment in queue (by priority) moves up

• If deleted experiment was next to run, following experiment starts when current experiment finishes

Priority Reassignment

Deleting an experiment does not automatically reassign priorities. If you want to fill the priority gap:

1. Identify experiments you want to move up

2. Manually adjust their priorities

3. Save changes

4. Queue reorders based on new priorities

Next Steps

• Experiment Status Lifecycle - Understanding status requirements for deletion

• Understanding the Backlog - How deletion affects your experiment queue

• Creating Experiments - Recreate deleted experiments with correct configuration

• Troubleshooting Experiments - Fix ERROR status experiments instead of deleting