Creating an A/B Test

A/B tests measure the success of messages and other areas of your app. Each test compares a control group of users, who receive no experience changes, to one or more test groups (also known as variants). In Leanplum, you can A/B test your app's variables, messages, and interfaces. There are two different ways to create A/B tests.

Method 1: From the Message Composer or Interface Center

If you are creating a new message or interface, you can create an A/B test directly from the Message Composer or Interface Builder.

Click the additional options button ( ... ) next to Send Now in the message composer. This will reveal an A/B test option in the dropdown. 


Choosing A/B test from the dropdown will create a new test based on the message or interface you were creating. Your message or interface's targets and other settings will automatically be applied to the A/B test. You'll have a chance to edit these settings and preview the test before you send it out to users.

Note: In general, it's best to test new messages and features before sending to all your users. If you decide to test a message or UI that is already active (or live), the test will only include users who trigger the message after the test start date.

Method 2: Test directly from the A/B Testing Dashboard

You can also create an A/B test directly from the A/B testing tab. Click the Create A/B Test button to get started.


Next, you'll be able to configure all your test settings, including your goals, test targets, variants, and more. See below for more on each step in setting up your test:

  1. Setting and reaching your goals
  2. Targeting your A/B test
  3. Using impression criteria
  4. Setting the distribution
  5. Add and edit content for each test group
  6. Estimating your test completion date and making changes
  7. Preview an A/B test on your device
  8. Gradually roll out A/B tests to your audience
  9. View A/B test results

Note that you'll need to register a test device to preview your A/B tests before starting them.

Didn't find what you were looking for? See the A/B testing section of our Help Center.

Was this article helpful?
Have more questions? Submit a request