Daniel Rowles’ A/B testing strategy is not all-encompassing.
While Daniel Rowles offers great advice regarding email marketing in Mobile Marketing: How Mobile Technology is Revolutionizing Marketing, Communications, and Advertising, he doesn’t supply a proper recommendation for A/B testing in an email journey containing more than two email variations. He suggests dividing 30% of your list into two, and A/B testing two different email variables, then sending the winning variable to the remaining 70% of your list (Rowles). If you’re sending out a one-off email to a certain subset of your users, this A/B testing strategy can work very well. However, if you are setting up a journey with emails that can be triggered by actions taken by a user, you may want to view A/B testing a bit differently.
When it comes to testing, the possibilities are infinite.
While A/B testing allows you to test two different variations, when running a test in a journey, the possibilities are infinite. Let’s use eCommerce as an example to illustrate what this can look like. Let’s say that you notice that a lot of your customers are abandoning their shopping carts and you want to remind them with the hopes that it will bring in more revenue. You design an email specifically for this list of users who have abandoned their shopping cart. You can start by A/B testing using Rowles method. Once you have the winning email that has the highest conversion rate from this initial test, name it your “control”.
You can expect that more users are going to abandon their shopping cart. You’ll want to continuously send out this email. The best use of your time and effort is to put the email into a journey that can automatically send it to a user who abandons their cart. Start by adding in your control. For each variation, change only one element of the control. Version 2 may have a different subject line. Version 3 may have a fresh design. Version 4 may have a different call to action, and so on and so forth.
The new winner becomes the control.
Split the volume of your users evenly across each email variation to start. Decide how long you want to run the test or how much volume you would like to see before declaring a winner. This new winner will become the control. You can decide based on conversions, which email you want to keep running and which emails to take out. Put higher volume on the emails that are performing the best, and lower volume on new emails until you can see how they are performing.
Never stop testing.
Create new versions of emails based on your new control, changing only one variable at a time. For Version 5, you may test the content. For version 6, consider another subject line. You may discover another email that performs even better. Test and repeat.
Rowles, R. (2017). Mobile Marketing: How Mobile Technology is Revolutionizing Marketing, Communications, and Advertising. Kogan Page Limited.