A/B Splitter
A/B testing empowers you to refine your AI conversations by comparing two variations and measuring their effectiveness. Imagine presenting users with slightly different conversation flows at random. A/B testing allows you to determine which variation drives users towards your desired outcome, such as completing a purchase or scheduling an appointment.
Setting Up A/B Testing:
Split Groups:
The Story Builder provides a slider to define the percentage of users who will encounter each variant.
For example, if you set Group A to 20% and Group B to 80%, approximately 20% of users will experience the chat flow associated with Group A, while the remaining 80% will interact with Group B's flow. This allows you to analyze results without exposing all users to potentially less effective variations.
Assign Blocks:
Next, you'll designate the specific blocks within your story that represent each variant. Enter the block name for Group A (e.g., "One") and Group B (e.g., "Two"). This links the user groups to their corresponding conversational flows. Note: Cards placed after A/B split will not run
Let's take a Example
We take 2 blocks
Block 1 named "BOX 1" with text "Yes" and Block 2 named "BOX 2" with text "No"
Now in another block we add a A/B Splitter in which firstly we'll assign "BOX 1" in Group A and "BOX 2" in Group B. Than we'll set the ratio to 10% to Group B and Group A with 90%. In the output we'll see that Group B will be given the preference and we'll get "No" as the result.
Now we set the ratio to 90% to Group B and Group A with 10% and we'll get the result as "No"
Last updated
