📘 This article explains creating and managing split tests with PageEditor optimizations. PageWorkers is part of Botify's Activation Suite, available as an option with all Botify plans.
Overview
When creating a Page Editor optimization in PageWorkers, you can indicate if you want to conduct a split test to help you evaluate your ideas before implementing them across your entire scope. Read more about PageWorkers split testing.
Creating a Split Test
Creating a split test in PageWorkers is the same process as creating a Page Editor optimization, with the additional step of enabling the split test option and selecting the distribution of pages to test.
To create a split test:
Create a Page Editor optimization applied to your entire website or a rule-based scope.
In the Split Test section of Step 1, slide the “Set up a split test” toggle to the right.
From the Variant Group dropdown list, select the percentage of pages in your scope to be used as the group that will receive the optimization during the split test. The Control Group percentage automatically updates to bring the total of both groups to 100%. We recommend keeping the default 50/50 allocation for best results.
Continue to define the optimization as with any Page Editor optimization and deploy.
The optimization is added to your list on the Overview page, labeled as a Split Test:
Monitoring a Split Test
Immediately after deploying a split test, reporting is available in the optimization details.
To access the report:
Click the name of the optimization on the Overview page.
Navigate to the Split Test tab.
The Split Test report includes the same charts as standard optimization reports segmented by variant and control groups, excluding the pages that receive the top 5% of clicks. Additionally, the split test status and KPIs are displayed at the top:
1: Summary of split test group distribution.
2: Current status. Read how to interpret results in About PageWorkers Split Testing.
3: How long the test has been running, and when the test was deployed. Pausing a split test has no impact on the day counter.
4: The number of clicks the variant group received per URL. If there is a difference between the variant and control groups, it is displayed below as a percentage of change.
5: The number of impressions the variant group received per URL. If there is a difference between the variant and control groups, it is displayed below as a percentage of change.
6: The variant group's click-through rate (CTR). If there is a difference between the variant and control groups, it is displayed below as a percentage of change.
Managing a Split Test
As with all PageWorkers optimizations, you can pause, edit, and delete split tests. Pausing a split test will not pause the counter in the split test report. Editing a split test will reset the counter and all reporting as if you created a new split test.
Ending a Split Test
You can end a split test anytime; however, we recommend running tests for at least 30 days for best results. If results are inconclusive after 30 days, we recommend continuing the test for another 30 days.
To end a split test, determine whether you want to deploy the optimization to the full scope identified in the optimization settings:
Deploy to Full Scope: This option is best when your split test produces a positive result. It will stop the split test and change the optimization from a split test to a standard optimization.
Do Not Deploy to Full Scope: This option is best when your split test produces a negative result. It will stop the split test and delete the optimization.
Deploy to Full Scope
Navigate to the optimization on the Overview page.
Select Edit from the context menu.
In the Split Test section, slide the toggle to the left, then click Next at the bottom of the page.
On the next page, click Deploy. The split test stops and is changed to a standard Page Editor optimization in Live status.
The split test report is deleted when you disable a split test.
Do Not Deploy to Full Scope
Select Delete from the context menu.
Click Yes, delete to confirm. The split test is stopped, and the optimization is permanently deleted.
See also: