You should test your email marketing messages just as you would test direct mail pieces. In the industry, this is called A/B testing. Basically, it means that you take a percentage of your email list, divide it in two, send different emails to each group and measure the performance of each to determine the most effective message.

While the process is fairly simple, there are a lot of variables. Here are some you should consider as while A/B testing your own email marketing.

What should you test?

There are many possibilities, by one tally, at least two dozen possible elements for A/B testing. The subject line is most commonly tested because of its impact on whether an email is opened. You can also test the “from” line; special offers; images; headlines; the message itself; the time of day or day of the week the message is sent. The list goes on. Start with the elements you consider most important. It is best to test one element at a time. Otherwise, it can be difficult to determine which factor affects the email’s effectiveness.

Avoid the spam folder

Spam can cripple your email marketing efforts. Test to ensure that your email is not being caught in spam filters. Some 20 percent of email messages are kicked into spam folders.

Bigger is better

The larger your sample, the more accurate the results. That said, small businesses typically have a relatively small email list so they must adjust their testing accordingly. The rule of thumb is 10 percent of your list, 5 percent for your A test, 5 percent for your B test.

Leave it to the professionals

Email campaign software has built-in tools for A/B testing, which makes it easy to test and to get quick and meaningful results. Businesses that use these tools are more likely to test their emails because it takes less time and effort.

Be thorough

Make sure you evaluate on several levels. If you want recipients to take a specific action, for example, sign up for a newsletter, be sure to measure that. Most businesses measure rates of opening, click-through and conversion, and you can set up A/B testing for each metric.