Navigate the A/B testing hub with ease

Learn how to masterfully navigate the A/B testing hub to easily get to an individual test, perform actions with a test, or select a group of tests for meta analysis.

The A/B testing hub allows you to keep track of all your tests and to perform meta-analyses on selections of them with ease. Here is how to make the most of it.

Key interface areas

The front page of the A/B testing hub is where you navigate to individual tests, filter or sort lists of tests for an overview of all your A/B testing activity, past and present.

key blocks of the A/B testing hub interface

The top of the interface contains the filters. They control everything displayed below them. The button for creating a new test is to the right, which will start the test creation wizard.

Below the filters there is a list of some key statistics such as the number of tests in the selection, the number of completed tests among them, their mean and median effect size and similar statistics for the winning tests only.

The statistics are followed by a graph displaying the confidence interval of each test in the selection. This gives you a quick overview of the performance of the selected A/B tests.

A list of tests is displayed below the graph. It contains all tests that match the filter criteria and is sorted by one of the columns being displayed. You can change the sorting by clicking on a column header. Clicking it again will reverse the sorting direction.

The buttons to perform a meta-analysis are located in key areas. Pressing them loads a meta-analysis of the currently selected tests.

A/B test list filters

The most useful control is the "Test categories" drop down which allows you to quickly filter to only tests of a certain status. The "Recently updated" and "Recently completed" categories should be of utmost interest.

Test statuses like "Active", "Completed", "Not started" can also be used to filter the test selection. "Not started" is any test which has a start date in the future. "Completed" is any test which has reached a conclusion either way.

Finally, you can filter tests based on their outcome by "Winning" and "Not winning" (only Completed tests are included here).

If you have defined projects while creating new tests or editing existing ones, these can be used to display tests belonging to a given project.

Tests can be filtered by name as well. Simply input a string that matches or partially matches the name of one or more tests, then click outside the field to trigger the filter.

Date ranges can be used to select only tests which have been active for at least one day during the selected date range.

Combining filters

All filters work in combination, meaning that for a test to be displayed in the list below and to participate in a meta-analysis, it has to match all the filter conditions. For example, if "Active" has been chosen from the test category drop-down, and "GO" has been entered as a text filter, only currently active tests that contain "GO" in their name will be displayed.

Similarly, if "Active" has been chosen and a date range has been specified, only tests which are currently active and which have been running for at least one day during the specified date range will be shown.

Actions with individual tests

In the last column of the test list two links will be present. The first has a small graph and a text label, such as "Analyze". This primary action will vary based on the current status of the test. Additionally, clicking the three dots will expand a drop-down menu with other possible actions:

A/B test actions drop-down

These may differ slightly based on the type of the test (single evaluation or sequential evaluation) as well as what kind of data source has been selected, or if a data source has been selected at all.

The drop-down can be useful if you want to quickly perform an action with a test, such as viewing its data, viewing or editing its data settings, statistical design information, or general test information. If using manual data entry, a quick link to enter new data will be present if adding more data is possible. Finally, you can also delete a test without even opening it through this drop-down menu.