There are a lot of tips regarding best practices, often not very good ones. At least once we’re all finished reading the reviews on the best color for the CTA: red, green or blue buttons? The truth is that the issue shouldn’t rely on the color itself, but on the balance and context in which is inserted.

So how do you know which color is better for your email campaign? Well, no one can tell you exactly.
That’s why the A/B Testing is so important.

Also called Split Testing, the A/B testing is a simple and automated method to check the performance of a campaign and if necessary to adjust the aim, achieving better results.

An A/B test is generally composed of two different versions of a message of the same campaign and, of course, a list of recipients. Currents ESP (Email Service Provider) have automated tools that send the two messages to two random sections of your database. After a set time, the ESP analyzes the performance index and determines the most effective message to send to the rest of your recipients.

How to make an A / B Testing?

Set your goal.
You cannot start an A / B Test without a goal. What results do you want to achieve? If your goal is to increase open rate, then you should focus on the object – date duo and sending time. If you aim for free download, then you’ll need to focus on the content and design of your email.

You can test many aspects of your email, such as:

  • Subject
  • Layout
  • CTA (Call to Action)
  • Headline
  • Copy and Tone of Voice
  • Buttons: colors, text, position
  • Offers and bonuses
  • Date and sending time

Use easy to analyze metrics

You should change one element of the messages to test, otherwise it will be difficult to know which factor made the email succesfully.

Set a deadline

How long does it take to decide which email is the best? It depends. If you are submitting a fixed-term offer, the testing time will be very limited and in proportion to its availability. In all the other cases, give to recipients the time to interact with the message. Based on the statistics of your previous posts, you should know the necessary feedback time.

Test and production environment must match

If your test is performed in the morning, even the selected email should be sent in the morning. Changing the global sending conditions does not ensure that the results will be the same!

Set up a percentage

If your test audience is made by a dozen users, then this becomes null. You should select a random sample of users between 10% and 15% of the total.

The test never ends

Although now you know better your target’s preferences and the best features to raise your open rare, don’t stop here. Continue for example to compare short against long subject newsletters.
Try to familiarize with each segment to which you address. What works for one may not work for another and likewise what works today may not work in a few months.

If you continue to test, you’ll continue to improve.