What is A / B testing about
The A / B test (also known as split test) is a way to improve our message by sending two slightly different versions of the email and using the version that achieves better results - depending on the assumptions, these can be conversions, clicks on links, openings email.
Examples of A / B tests
To clear up any doubts, here are two examples of A / B testing. I hope they will dispel all your doubts.
Which e-mail title should you choose?
John creates belly dance courses online and has already prepared a mailing, but he has two ideas for the title. Unfortunately, he can't make up his mind, so he decides to do an A / B test. Sends an email with the same content but with different titles. Up to two different groups each. 10 people each. In our example, we'll assume that he cares about the number of email openings (CTR) the most.
Two days later, he checks the results and they are as follows:
- "Belly dance - course in 10 days" - open by 3 recipients.
- "Show him your exotic side - master belly dancing in 10 days" - open by 7 audiences.
Fortunately, Adam studied mathematics for 8 years and after a bunch of integrals he knows ... 7 is greater than 3. The mailing recipients have voted - the second title is better. Adam sends an email with this title to the remaining 200 recipients.
What kind of graphics should I choose?
Julia runs an online hardware store and knows her audience well. She knows drill operators do not like to overpay for their equipment. She also knows well that CTA is one of the most important elements of mailing and that is why she prepared two graphics encouraging to buy:
- Red and yellow graphics in the style of MediaMark advertisements, with a photo of the drill and the words "Lowest prices - we do not screw it"
- Graphics with black text on a yellow background "Click here to buy cheap drills".
Julia does not know which option will be better, so like John, she decides to conduct A / B tests. In her case, the title of the e-mail and its content will be identical in both cases. The only thing that will distinguish these e-mails is the said graphic at the very bottom of the e-mail.
Julia is good when it comes to percentages, so she sends mail to 10% of her recipients - 5% for each graphic. She checks the results after a week and it turns out that graphic no. 2 won the test. He sends this graphic to other recipients.
Perhaps your case does not fit into any of the above examples, but it does not matter - using A / B tests you can check every element of your email:
- graphics used in e-mail
- layout in the e-mail
- the salutation you use
- e-mail footer
- many many more…
And how to make it easier?
I have to admit something here ... I made up the story of John and Julia, and the biggest lie is that John and Julia carried out these tests on their own. In fact, it's much better. Most mailing platforms have the option to do it for us. Our only task is to prepare the e-mail and its alternative version, which differs from the original by one element.
In short, the procedure for most platforms is as follows:
- Go to your platform for sending mailing and select A / B campaign to send.
- Prepare an e-mail and two versions of the tested element.
- Specify how many people the test e-mail should be sent to and after what time the test is to be resolved.
- Send the test and enjoy the automation - the platform will check the test results itself after a certain time and send the winning version of the e-mail to other recipients.
I could finish today's post with this algorithm, but there is one more thing left. An important issue…
More complex tests
In addition to imaginary stories, I will admit something else to you. There was a certain… understatement between us… Until now, we only dealt with the test area with one variable (eg title) and two variants. But what if we have more titles, or if we would like to test both titles and graphics at the same time?
In fact, there can be an infinite number of variants - it could be an A / B / C / D test. And you can combine the variables with each other - in one test you can check the title + footer. The latter example is the more powerful brother of A / B tests - the multivariate test.
There are endless possibilities for testing - the only barrier is your imagination and the limitations of the mailing platform you use. And if your platform does not support your dream A / B tests, remember that there are more of them on the market and maybe it is worth testing a new one? Of course, you can choose the best one with the help of our comparison tool HERE.
In the meantime, I wish you spectacular A / B tests and as always ...