Despite the word “test” being in the title, A/B testing is meant to be used to send real messages to your target audience so be sure to only use polished, ready-to-send emails with the tool.
Make sure you follow the pre-send checklist before sending to make sure your email and data are ready.
A/B testing only works for single send emails; it doesn’t work for automated messages in automation studio or Journey Builder. You’ll need to use other techniques for A/B testing automated messages
How to send an A/B test email
Log into Marketing Cloud
Navigate to Email Studio
Select A/B Testing from the top menu
Click “Create A/B Test” button in the upper right corner
Under “Test Management”, complete these steps
Name the test (this is the name of the test itself, and how it’ll show up in analytics, not the name of the email)
Provide a brief description saying what this test is
Select the Test Type (follow the directions for the type of test you want to do and then skip to the step 6)
Subject Lines
Choose the email
Check that your preheader is as expected
Paste or write in the two subject line versions
Click the “Next” button
Emails
Select the two emails you’d like to test
Select your desired subject line and preheader
Note: it’s best to minimize the number of variables you’re testing so you may want to use the same subject line and preheader for both emails if appropriate
Click the “Next” button
Content Areas
Before setting up the test: drag an “A/B Test” block into your email from “Advanced Content” in the Marketing Cloud blocks. Select saved versions of Email Framework content blocks you’ve added your version A and B content to.
Choose the email you want to test that has the A/B content block in it
Click the “Next” button
From names
Select the email you’d like to test
Select From Name A and From Name B
When using the email framework, the brand configuration in the email name will control the branding inside the email and reply to email address and name
However, you can't select the brand configuration for the from names A or B. Instead, you have to create a sender profile for from names A and B (you don't need to complete a delivery profile or send classification—that's controlled by the brand configuration).
Click the “Next” button
Send dates/times
Choose the email you want to test
Select the two different send times and dates
Click the “Next” button
Preheaders
Choose the email
Paste or write in the two preheader versions in “Preheader A” and “Preheader B” (you can leave the “Preheader” directly below the subject line blank)
Click the “Next” button
Complete the Recipients tab
Choose your recipients from a Data Extension
Note: You can only use one Data Extension at a time in an A/B test
Select your publication list. This can be the same publication list you normally use for a send of this type.
Select Exclusion and Suppression lists as needed
Select the test size
If you have more than 50,000 subscribers, we recommend sending to 5% for A and 5% for B. If you have less than 50,000 subscribers, we recommend using 10% for each.
The estimated number for each group does not factor in suppression or unsubscribes in the total
Even if you select the same percentage for each group, the number may be slightly different. This is okay.
There is not a way to assign specific subscribers to A or B. The system randomly assigns the test version.
Click the “Next” button
Complete the Winner tab
Select “higher unique open rate” or “higher unique click through rate”
Note: picking the “best” metric will depend on the specific test. Our test type chart provides suggestions for which to pick.
Select the “evaluation period”
Most email engagement happens within a few days of the send. If you have the time, select 24-48 hours for evaluating. If not, we recommend waiting at least 4 hours.
Click the “Next” button
Complete the Send Management tab
Start test
You can choose to start the test immediately (first option) or schedule it to send at a specific start date/time (second option). If you select the third option, “schedule later”, you will have to go back to the test to schedule it to send; otherwise, the email(s) will not go out; typically we don’t recommend this option.
You can also choose to send the winner to the remaining subscribers automatically or manually. Typically, automatically will be an easier, more efficient process. If you select manually, you will have to go back into the A/B test after the winner is declared to send that winner to the remaining subscribers.
If you select automatically, you can have the winner send immediately once the winner is declared or at a specific time
From options
Select the send classification you’d like to use for the send. For Email Framework sends, this will be “Email Framework - Commercial” or “Email Framework - Transactional”
Note: When A/B testing from names you’ll still need to select this but make sure not to override the delivery or sender profile
Send options
Leave as default
Email notifications
Include your email address and anyone else who should be notified
It’s recommended that you select all of the notifications
Click the “Finish” button
Note: Don’t make any changes to the data extension that you used until the test is finished
Evaluate your send
You should receive email notifications with the updates of the send status as well as general analytics
You can also access full analytics in the A/B testing folder under the Tracking Sends tab within Email Studio
Salesforce Marketing Cloud User Guide social media channels
