This is an old revision of the document!
Email A/B Testing
If there are competing opinions as to the best way to make your email attractive to recipients, or you are unsure about the best approach to take, then you can make a 'Split Test' or 'A/B Test'.
You pick 2 Emails and Sentori will send each to a sample of your selected audience. Once interactions with these tests have begun Sentori will make an assessment of which one is best performing. This best performing or 'winning' email will then be sent to the remainder of the audience.
What To Do
Prepare the emails
Set up the test
Take the 'Automation' tab from the main navigation and select 'Email A/B Testing'.
Create a new A/B Test. You can give the test a name and add a description if appropriate.
You will be asked to select the two emails to be used. In the event of a tie between the emails, email A will be considered the winner.
Select the contacts to receive the emails. This is done in exactly the same way as you would for sending a single email.
You now set the parameters for the test. These are the total sample size and the length of time to wait before analysing the results and sending the winner to the remaining contacts.
You select the sample size using the slider. Setting the slider to 20% will mean that each email will be sent to 10% of the total number of selected contacts.
Clearly Sentori has to wait a reasonable time before making the decision as to which has performed best. We would recommend waiting at least 4 hours.
Finally you queue the email and set the time that the test should start.
Frequently Asked Questions
How is the Winner Determined?
The winning email is decided by looking at the number of emails opened and also the number of interactions with the email. The two emails are given a score. One point is awarded for each contact to email open the email, one point for each person to click a link, and one point for each social share.
If the score for each email is the same, email A is considered the winner.
Is the Test Group Randomly Selected?
Yes, the two test groups are randomly selected.