A/B testing
Branching Experiments
A/B testing is a widget flow node that allows users to be randomly assigned to groups for the purpose of conducting A/B testing (also known as split testing).
Description
The widget flow node randomly selects a branch to continue the flow based on the specified distribution. The flow then proceeds along the selected branch.
Configuration
Distribution
Distribution – you can define how users are distributed across groups. Users will be randomly assigned based on the specified ratio. This widget flow node provides purely random distribution. If you require distribution based on specific conditions, consider using it in combination with the condition node.
Sticky
Sticky – you can enable the browser to remember the assigned group, ensuring that the same user (device, browser) consistently follows the same flow.
WARNING
The sticky key is based on the distribution configuration. This means that the sticky assignment will be reset if the distribution ratio is changed.
Limitations
| Name | Value |
|---|---|
| Min percentage step | 1% |
| Min groups | 2 |
| Max groups | 5 |
| Query parameter name max length | 32 |
| Origin pattern max length | 64 |
| Inputs | Exactly one |
| Outputs | 2 to 5 |
Context values
A group will be added to the context values, indicating the assigned distribution group. Its value will be one of the following letters: A, B, C, D, or E.