Manage variations and traffic allocation

Variations represent the experiences compared by an experiment. The control is the baseline used to interpret the alternatives.

Add and describe variations

1. Open the experiment detail page and locate Variations.
2. Use the add action, enter a variation name, and save.
3. Review the control and alternatives so each name matches the intended experience.
4. Adjust the traffic allocation from the variation list.

A newly added variation starts at 0% traffic. Adding it does not automatically change your testing platform or include it in a running test.

Distribute traffic

Use the allocation controls to set percentages. Lock a variation's allocation when you want it held while distributing the remainder. Distribute equally distributes the available traffic across eligible unlocked variations.

Confirm the planned allocation is valid for the full experiment and agrees with the provider configuration. The expected allocation is used for Sample Ratio Mismatch checks; it should describe the actual design rather than being changed to fit observed traffic after the test.

Map provider variations

A linked indicator shows a variation connected to a provider. Use Map for an unmapped variation and select the corresponding provider variation. Use the unmap action when a link is wrong, then select the correct replacement.

Supported analysis requires every provider variation to be mapped one-to-one, with a valid baseline and primary metric. Similar names alone do not confirm a correct mapping.

Deletion removes a variation and may leave a report without a required mapping or baseline. Review the experiment and result status after changing its variation set.