No one has perfected email marketing. If they have, I hope they are enjoying retirement on a beach and have the urge to pass along their knowledge to the rest of us. But until that box of secrets shows up at our doors, the burden is on us to stay ahead of the ever-evolving preferences of our audiences.

 

Email marketing isn’t stationary, and what worked last week may not work this week. In truth, it is nearly impossible for an email marketer to know what always works all the time. But, that shouldn’t stop us from learning! And the best way to learn? A/B test.

 

Let’s look at how to build an A/B testing plan for your email marketing campaigns.

 

Keys to A/B Testing

As mentioned in a previous blog, A/B testing must be two things—scientific and significant. I’m not talking about all 7 steps of the Scientific Method, but you should focus on a few of them. Be sure that you:

  • truly randomized your list segments
  • select large enough segments
  • only test one part of your email at a time

 

With each test, start by randomizing your list segments. You should do this step with every email you want to A/B test. Otherwise, if you continue to use the same Segment A and Segment B with each test, or decide to test both time of day and day of week, you will get skewed results.

 

You can randomize manually, or if you’re using a marketing automation platform, you can see if your tool has an option for facilitating A/B testing. If you’re using the emfluence Marketing Platform, for example, you can use the A/B…Z Split segmentation tool to re-segment your list with each send.

 

You don’t always have to do a 50/50 test either. If you’re testing subject lines, I suggest a 10/10/80 split. In this case, you would send different subject lines to two 10% segments. Wait to see which has more opens and send to the remaining 80%. A general rule of thumb is to have a large list to allow for A/B testing. Some will suggest about 10K records per segment. I realize not everyone has that large of a list, so I think that if a segment has a few thousand records, you will see some actionable results. If they’re too close to determine a difference between the two elements – test again! Also, be sure to schedule your tests so they deploy at the same time to prevent skewed results, unless of course you’re testing time of day!

 

There’s also the significant aspect of A/B testing. You are a digital marketer who is used to looking at data to determine how you approach the next campaign. Why treat email any differently? The results that you are finding about your A/B next need to be actionable, and meaningful. How are you going to use the results to change the next email? Maybe Friday does work better for your audience rather than a Wednesday. Great! Send the next email on Friday. Don’t just A/B test because your account manager has been bugging you about testing…do it to learn about your audience and improve your engagement rates.

 

What Can I Test?

Your opportunities for testing are essentially endless. To get the thoughts flowing, consider testing any of the following:

 

  • Time of Day
  • Day of Week
  • Subject Lines or Pre-Header Text
  • From Name
  • Images Heavy vs. Text Heavy
  • Personalization
  • CTA Button Verbiage

 

 

How Do I Read My Data?

First off, it’s awesome that you decided to dip your toes into the water of A/B testing…but how do you read the results? Most likely, you will look at views or clicks, but knowing which email was successful is about determining which metrics to compare. If you have tested an element of the email that would entice the reader to OPEN the email, you will look at Unique Views – Time of Day, Day of Week, From Name, Subject Line/Pre-Header. If you are seeing how the reader ENGAGED with your email, you will want to look at the Unique Clicks. This includes items such as the personalization, email format, CTA verbiage, links, email tones.

 

Not every A/B test is going to have dramatically different results between the items being tested. But that doesn’t mean that your testing has failed, it just means you need to continue to test to find what works with your audience.

 

Need help with you’re A/B testing? Feel free to reach out to your account manager or support@emfluence.com.


Leave a Reply

Your email address will not be published. Required fields are marked *

 

Let's Get Started