Secrets of Using Email A/B Testing

To find the best recipe for success with your email newsletters, experiment and adjust. Commonly called A/B testing, the process of finding out what your customers respond to requires patience and time thanks to the number of variables at play.

Secrets of Using Email A/B Testing

Let’s start this discussion with a pop quiz about email marketing.

Would you rather have 25 percent of your prospects open your emails or 8.5 percent engage with them with a click or two?

The answer is both. Or neither. It depends on your objectives. If you sell via the internet, you may be more concerned with how many people click on a buying link and then how many of those follow through with a purchase transaction. If you’re looking for mind share to encourage readers to swing by the store, the number of prospects who “see” your email message might mean more to you. 

Once you’ve decided what you want to optimize, you then must tweak your emails to make sure they’re performing optimally. For example, if your choice in the above pop quiz was to maximize engagement, you’ll want to experiment with many email components to see what works best. What happens if you change the subject line of your emails? What about the style of the body content? Should you use images or text only? How long should your emails be? What day of the week should you send? What time of day?

These factors, and others, can impact how many people open, read and act upon your email messages. To find the best recipe for success, you must experiment and adjust. Commonly called A/B testing, the process of finding out what your customers respond to requires patience and time thanks to the number of variables at play. 

What is A/B Testing?

Instead of guessing which types of emails your customers are more likely to open, the process of A/B testing tries multiple variants to see which works best. Email services normally build this functionality into their systems, so you can easily create two or more versions of the same email and send each to a percentage of your overall email list. Results are tracked for each so you can see performance metrics such as open rates, click rates and more. 

For example, the simplest type of A/B test might involve testing two different subject lines for the same email. One email might have a subject line such as, “Check out our huge selection of long-range optics.” The other might stick to the same theme, but present a different hook for the reader, perhaps, “Learn why first-focal plane scopes might be right for you.” Both offer a call to action. The first emphasizes shopping selection while the second wraps the scope sale message into a learning promise. To find out which will engage your customers you must test both. 

While simple A/B testing might just test subject lines, the process can be applied to everything about the email messages including subject, content, look and feel, tone and more. A/B testing can also be used to test performance attributes other than open and click rates. For example, you might use the method to see which types of emails do better at avoiding spam filters. More on that later. 

Also keep in mind that true A/B testing doesn’t define your limits for testing in general. While most email systems allow you to split a single campaign into test segments, you can do your own testing with other variables over time. 

What Are Variables To Test?

— Subject Line

The most obvious thing you can change is the subject line. In a typical email reader client such as Outlook, Mac Mail or Gmail, the viewer will see a long list of emails in a preview mode. It’s not uncommon for people to get 50, 100 or even more emails per day, so your title has to stand out.

When choosing and testing titles, don’t forget the visual information you get for free. For example, most every email reader will show the “From” address, so when setting up your email system, choose the sending name wisely. For example, if your emails show up as sent by “ACME Tactical Supply” then you don’t need to waste one precious character of subject line space repeating your business name. So, “Weekly update from ACME Tactical” will be redundant. Instead, rely on the from address to cover your branding and use the subject line as a call to action. For example, something like “New Trijicon SRO Sights Are Here!” may get you more mileage as people already see your business name in the from line. 

When you go into testing mode, start with subject line alternatives and don’t be lazy. Remember, this is the key variable that encourages people to open your email. With average open rates ranging from 10 to 30 percent, making the subject line work for you is important. Also remember that your opinion on subject line doesn’t really matter. If “Get 20 Percent Off Until Wednesday Only” tests better than “Best Gear at the Best Prices” don’t take it personally. Go with what your customers choose with their clicks. 

— Preview Line

The preview of most email clients will show several pieces of information that reader can evaluate before deciding to open a message. Date, sender name and subject are a given. Most email readers also display a preview line or two that give a hint to what’s inside the message. Many email systems allow you to customize this content, so be sure to test compelling preview messages too.

— From Address

Even the “from” address can have an impact on email performance, so test that too. Are customers more likely to open an email from or Testing will tell you. 

— Day Of The Week

Typically, the best days of the week to send most types of emails are Tuesday through Thursday. However, as a retail business where prospective customers may be more interested in catching up on their off days, you might try weekend deliveries too. For several email lists I run in the guns and shooting industry space, Saturday deliveries aren’t statistically different from mind-week sends for open and click rates. 

— Time Of Day

There’s only one way to determine when most your prospective customers read their email. Test. While not part of true A/B testing, it’s easy enough to try sending your mails at different times of day over a month or so to see which gets the best response.

— Sending Frequency

Your sending frequency is also a variable that can impact reader engagement. When testing this variable, look at how more or less frequent emails from you impact not just open and click rates, but unsubscribe and abuse reports. If you send your emails out too infrequently, readers may forget that they subscribed in the first place and report your messages as spam. Too frequent and you run the same risk. 

— Body Content Length And Format

Be sure to try different email formats over time. Why? The average attention span of a human is shorter than that of a goldfish. That’s a real statistic. People will focus on one thing, such as scanning an email, for just eight seconds. So, try making your emails easily scannable. You can experiment with photos and titles only, or titles and short descriptive text about what’s behind the headline. 

Testing For What?

It’s important to identify your desired output before embarking on test campaigns.

Remember that testing requires patience and lots of time. To get any meaningful results, you must establish a baseline over three issues for each variable you intend to change and track. For example, to know your open rates for one type of subject line style, you’ll want to get results for three different sends and average them.

Each time you make a change to one variable, you must run at least three more issues to see the results. As email marketing evolves, you’ll want to keep track of performance and continue to optimize forever.

Open Rate

Nothing else matters unless potential readers open your email, so this is the first metric you’ll want to develop. 

Email sending programs and services include a small invisible graphic link in outbound emails for determining if an email was opened. If the email is filed or deleted without being opened, the tracking link never “calls” its server for that invisible image and we can assume that the prospective reader didn’t open the message. If the message is opened, the link functions, and an “open” is recorded in the tracking stats. It’s not an exact science, but open tracking works pretty well. For trend analysis over time, its precision is good enough. 

Open rates vary by industry and type of email campaign, but for general retail, a rate of 19 percent is average. The rate for your business can vary in either direction, so use this figure as a starting guideline. The important consideration is whether you can increase your rate over time by testing and tweaking email content and design.

Click Rate

When you include a link in an outbound email, your mail provider will automatically modify that link so the click passes through their server first. For example, if you include a link such as, the final email might translate that link to something like, or a coded equivalent. The reader is ultimately passed to your site, transparently to them. However, since they passed through your email provider’s server on the way, that activity can be tracked. 

Again, for general retail businesses, the industry average is about 2.5 percent. As you can see, email marketing and promotion is a numbers game as four of five people won’t open it and over 95 out of a hundred won’t click on anything. Don’t get discouraged, however, with good testing and optimization, you can improve your results significantly. 

Spam Trap Avoidance

One testing variable not often considered is spam filter evasion. While many email systems offer spam checkers that help you analyze the likelihood of your email getting flagged as spam, this only provides a guideline. You won’t know for sure what percentage of your emails are getting through to users until you actually send them. To illustrate the challenge, I’ll share a personal example. 

As part of my job, I subscribe voluntarily many different email newsletters and updates. One of my email accounts is a address provided by Apple. I use the built-in Apple Mail program as my reader. Both the service and the Mail app have spam filtering functionality and both often make independent decisions about what is and isn’t true spam. 

Things get really interesting because spam decisions aren’t always made by considering the sender, but by individual email. As an example, I subscribe to the National Shooting Sports Foundation updates. To be clear, I choose to receive emails from this sender, so I never mark them as spam. Some emails from are delivered straight to my inbox. Others are directed to the spam folder. Why? Got me. Even though I regularly mark emails as “not spam” they still get caught in the service-level filter. 

So, what’s an email sender to do?

Unfortunately, making sure your emails actually get delivered to people who asked to receive them is a never-ending task. You can’t set up once and assume everything continues going to plan. While you’ll be able to see which emails were manually flagged as SPAM by your readers, you won’t be able to get a simple report that shows you how many of your emails were marked as spam by the service itself. The very last thing those in the spam filtering business want to do is help people they consider (even wrongly) spammers to get their emails through. One indirect method of gauging deliverability is to look at open rates. Assuming you send similar types of emails, your open rates will fluctuate within a known range. 

Unsubscribe Rate

If you send marketing emails, you need to include an unsubscribe link. In fact, most commercial email services require such a link and, in some cases, automatically add one. One measurement you’ll want to keep an eye on as you experiment with A/B testing is the number of unsubscribes. If a testing change causes a spike in the number of unsubscribes, you’ll know that something you’ve done is decreasing the perceived value of your message. 

Abuse Rate

Some email clients, such as Gmail and Yahoo, offer users a “Report Spam” button. This is a helpful tool for reporting actual spam but, unfortunately, many users hit this button in place of an unsubscribe. Yes, someone who voluntarily added themselves to your email list may report you as a spammer and impact your reputation and ability to deliver emails to other customers. While you can’t eliminate this unfair behavior by some readers, you can minimize it by monitoring the complaint rate and being crystal clear in your emails and signup forms that users can unsubscribe at any time. 

Parting Shots

E-mail marketing presents some unique challenges and opportunities. It can be a powerful business development tool but it’s also an ever-changing complex minefield. Not only do you have to worry about how multiple variables work together to define success or failure, the rules of the road are always changing. Imagine a highway where lanes appear and disappear, hairpin turns materialize out of nowhere and dead ends can end your drive without notice, and you’ve got the idea. 

The good news is that email services today can help you navigate the potholes so you can focus on your customer message. If you want to be industrious, dive into the details of email management and delivery, and pay a lot less, you can use a bare-bones a service such as Amazon SES. On the other hand, there are plenty of email providers that worry about details like delivery rates and white-list maintenance — you’ll just pay more for that peace of mind. Whatever route you choose, you’ll need to understand the importance of testing your email campaigns to find out which combinations of variables work best for the objectives you plan to achieve. 


Comments on this site are submitted by users and are not endorsed by nor do they reflect the views or opinions of COLE Publishing, Inc. Comments are moderated before being posted.