All Collections
Split Testing
How to set up an Audience Split Testing experiment
How to set up an Audience Split Testing experiment
Apoorva Wate avatar
Written by Apoorva Wate
Updated over a week ago

To begin creating a split testing experiment:

  1. Navigate to the Experiments page from the Nav bar and click on the “Create Experiment” button.

  2. Select “Split Testing” under Experiment Type

  3. Select the date range for the experiment, ie the from which you want to start the experiment and when it should end. It should be a minimum of 7 day period. There is also an option to end the experiment when statistical significance (95%) is achieved.

  4. Select the audience you want to experiment with. You can choose any existing segment or list from your workspace. The audience should be of at least 1000 contacts to experiment.

  5. Select the metric (KPI) whose lift you want to measure. You can measure the following metrics:

    • Store Visits (count of page_view, product_view, product_list_view)

    • Cart additions

    • Revenue

    • Number of orders

    • Check-outs (count of checkout_create)

  6. Next, create the Splits for the experiment. With Splits, you can create randomized groups of users within your audience. You can specify:

    • Control group - Group of contacts that will not be exposed to any campaign.

    • The number of treatment groups you want to use.

    • The percentage of total contacts that each split should have; for example, Control 10%, Treatment A gets 45% and Treatment B gets 45%

    • The destination to send each Treatment to; for example, Treatment A goes to Facebook and Treatment B goes to Google. The marketing channels that you have integrated with Lifesight will be available for selection. Contact of Control isn’t sent to any destination.

  7. Name the experiment and click on ‘Create Model’ to proceed.

  8. Once the experiment is created, you will be able to view it on the Experiments page. The status would be represented according to the start time set by you.

  • There would be only one control group.

  • Lifesight allows up to four treatment groups to be added as a Split.

  • All treatment groups would be of the same size.

  • Users can select auto-split, and the contact will be equally split among control and treatment groups. If auto-split is not selected, the user will have to input the percentage of the total audience membership to any of the treatment groups and control groups, and the rest of the percentage will be filled with criteria of equal treatment groups and remaining in the control group.

  • An ad platform can be selected multiple times as a destination.

In case you have any further queries, feel free to write to us at [email protected] and we’ll respond at the earliest.

Did this answer your question?