Split Test for Success

There’s a couple of topics that have generating a lot of discussion of late. Deliverability is one; another is how to continue to grow your opt-in email list. With so much talk about these topics, one wonders how much time and effort is being spent on optimizing the content of your email messages now that you’ve got a sizable opt-in list and have overcome your deliverability challenges.

One very underutilized tool in email marketing and campaign development is the Split Test. Most email solutions provide some type of split test function (usually called “A/B Split Test,” but often with the ability to run a test of more than just two components), but only a very small percentage of marketers actually use it—we estimate less than ten percent, possibly even less than five percent. This is a shame because a well-designed split test can help you take the content that you’ve developed and truly hone your campaign into one that will generate good results. And it’s easy to do!

So, how does one go about running a split test? It’s simple. Take the content from your next email campaign—the subject line, text, offer, images, etc.—and develop various alternatives. For example, you might test the same content, but with different subject lines. Or you might keep the subject lines and content the same, but include a different promotional offer in the email. You might even provide different images to see if those improve the response. The key to a meaningful split test is to conduct the test by changing only one thing in each different version. If you run a test where you change both the subject line and the content, and one version in that test performs better, how do you know if it was the different subject line or different content that made the difference?

Most email programs enable you to create your tests quite easily. In ListManager, for example, you simply choose either a percentage or an absolute number of recipients to which you want to send the test versions, and ListManager automatically selects a random subset of the list based on this figure. Then, you load the different test emails, as many as ten, and ListManager sends the different versions to this subset. If your subset is 40 names and you have four different versions, each version will be sent to ten different names.   

After the emails are sent, you view the results to decide which version performed the best. You can base your assessment on any factor that you determine to be significant: open, clickthrough, or conversion rate. Once you determine which test version was the best, then you send that version to the rest of your list (again, ListManager can do this automatically) and sit back and see how much improved the performance of this campaign was over the campaigns in which you did not test different variations beforehand.

This is what makes a good marketer into a great marketer. Utilizing the tools at your disposal to continually hone and improve your marketing efforts so you can be assured your providing just the right content to get just the results you seek.

Contact us at editor@lyris.com to share your ideas. We may include it in the next issue of Making Mail Work!