If you were to Google “email marketing best practices,” you would be presented with a long list of do’s and don’ts regarding every asset of sending an email. I could certainly write a whole series on spam words to avoid, best times of day to send emails, how to write a subject line, and many more. But even then, who is to say that your prospects aren’t nocturnal? That they don’t like super long subject lines, or that some email client in Vanuatu loves the word “free” in emails and lets them into inboxes, a-hem, freely?
When your bottom line sales depend on how much traffic you get through the top of the funnel, there’s only one way to ensure you are maximizing your specific prospect base and pushing as many through the funnel as possible, and that’s Testing.
Commonly referred to as A/B Testing, testing refers to splitting your prospect base into sections, testing different variations of emails, landing pages, ads, etc. on test groups, and then presenting the best-performing asset to the rest of the database. I’m addressing email testing specifically in this post.
Many of my peers “him and haw” when they think of having to create multiple emails, divide a list and wait to see results; They think “I’ll do it next time when I have more time.” Here’s a little secret…you’re never going to have more time! If it means starting a promotion a day late or pushing back an email, it’s worth it.
So how do you set up A/B testing?
Think way, waaaaaay back (well, maybe farther back for some of us than others) to 7thgrade when you had to do that science project with plants and what kinds of light encouraged the most plant growth? This is kinda like that. You have a control group and a test group to test your email design. Generally, you will divide your prospect list into 3 parts — control group (10%), test group (10%) and the remaining 80%. This last section will be sent the winning email once the test has been ran. If you’re unsure how to do divide your database, contact your email solution’s support team.
**Please note: If you have a large database (1 million contacts), you may want to use two 5% parts for your control and test groups instead of 10% parts. These groups need only be statistically significant and remember, you want the winning email going to the largest majority.
Once your list is divided, design the emails. For the control group, design an email similar to those you’ve sent in the past (Email A). For the test group, change ONE element of the control email to see if it makes an impact on engagement (Email B). There is an endless list of email attributes you can test, but here’s a list to get you started:
- Subject line
- Personalization (Starting the email with Dear prospect name, v. generic salutation)
- Placement of call-to-action (CTA) button
- Color of CTA button
- Text on CTA button
- Graphic v. text email
For more ideas, visit Anne Holland’s Which Test Won.
Once your emails are designed, run the test. Send Email A to the control group and Email B to the test group. Wait at least one day to tabulate results. Once you’ve determined which email outperformed the other, send the better-performing email to the remainder of the list.
Rinse, repeat. You’re never done testing and you’ve never got it all figured out. Mastering the art of email marketing is not a matter of “having the magic touch.” It is a matter of testing, testing, and testing some more, and being able to make business sense of the results and implementing them in a way that makes an impact on your bottom line.
For more information on email marketing and multivariate testing, feel free toemail me, or message me on Twitter (@jennybeal).