Email testing and optimization: an in-depth look into what works

ryan-phelan-headshotGuest column by Ryan Phelan, Adestra.

I’ve been in the email marketing industry for fifteen years, and have seen firsthand what works and even what doesn’t work in email testing and optimization programs.

So here are some tips and tactics from my email marketing experience to help you in your testing efforts.

Headlines, headlines, headlines

Subject lines are always popular email items to test and for good reason – the subject line and the From field are the first two things an email recipient will see with your email send.

There’s plenty of information and misinformation out there on what words and phrases are effective or ineffective and what phrasing can result in getting your mail sent immediately to a spam or junk folder.

This chart is from a recent study that outlines the top 10 and bottom 10 performing subject lines taken from more than three billion email sends:

guest-ryan-keword-chartBut while testing subject lines is important, you shouldn’t put all your testing eggs in the subject line basket. If you want to improve clicks and engagement, your go-to test should be other email elements, such as copy or design.

When testing any aspect of your digital marketing strategy, it’s imperative to strive for statistical confidence in the results.

The challenging aspect of conducting a statistically valid test is that it requires around 40,000 names in each segment – think different subject lines, for example – to be tested. Therefore, in order to run a valid test comparing subject line “A” to subject line “B,” you’ll need a list of at least 80,000 names.

probability-flickr-3676624894-ee4142a611-z

Add two headlines to that overall test in order to find out what creates the best open and click-through rates, and it becomes a four-way multivariate test that needs an audience of at least 160,000 names to be effective.

Finding a large testing audience might not be a problem for marketers that already have large lists, but running complex tests like this might pose a real challenge for marketers with a smaller subscriber list.

If you try to run the tests sequentially – say testing subject lines one day, headlines another day, then mixing it up with subject line/headline combinations for two more days – it’s important to introduce a completely new variable in the process because each test was run on a different day. Therefore, you won’t be able to isolate the change in the results with the same certainty.

This reality means marketers need to be honest with themselves when planning the complexity of tests they want to run, and be willing to get back to basics and focus on list building if needed.

Organizing your testing program

One problem I see all too often is marketers not keeping track of the tests they run.

organize-information

Marketers should have a binder or a spreadsheet that can be as simple as tracking five basic elements:

  1. Information on what was tested
  2. Information on why it was tested
  3. The results of the test
  4. The plan of action taken as a result of the test
  5. The revenue from the test

To get more specific, the spreadsheet could also include a screenshot of the email tested and links to the asset library to visually keep a record of the creative elements of the campaign.

You should keep your test tracking internal to the testing team. Then at the end of every quarter, you can put together a report that can be presented to management and recaps what the testing accomplished, what was learned, and what was improved through the testing.

Along with keeping a record of what was tested, your program should also involve a three-month rolling plan on what to test based on the results of previous tests. This will help organize the process, rather than finding your team testing random elements of different email campaigns.

Remember testing is an ongoing process and not an end goal

I always advise people start at the beginning of the quarter and map-out what your tests are going to look like at every send. You’re trying to get from A to Z, and with this in mind, figure out what you ultimately want Z to look like and determine what action needs to be taken to get there.

For instance, in order to prove that a particular header works, you have to test all iterations of this header. Running a series of one-off tests won’t prove anything. Rather, it only shows that the one format being tested only works for one particular message for that specific date and time, and thus it is not replicable.

All of these factors, from the meticulous details to the big-picture endgame goals, must be digested and considered. This is why exercising proper testing from morning send to evening send, within every relevant demographic, segment, and more could take anywhere between a month and three months.

The most important thing is to just get started. Remember that “patience is a virtue” and email testing and optimization is a journey and not a destination.

Ryan Phelan is the vice president of marketing insights for Adestra.

Article images via Flickr.

Related reading

Email
Email marketing
email notification
inmail-blog-image
<