Blog | May 22, 2015

How to A/B Test Your Way to Success

by Emily Beaver in Marketing Operations

Want to find out which call-to-action button will get the most clicks or which headline will get the most engagement?

Like many marketers, you might decide to set up an A/B test to find the better-performing headline or button.

A/B testing, or “split testing,”  has become popular among marketers for good reason. The technique, which presents two possible versions of content or designs to different groups to measure which performs better, is becoming more accessible. There’s a plethora of comprehensive guides on A/B testing and online tools to help marketers optimize their marketing strategies.

Before you launch a test, it’s worth thinking about how to get the most out of your efforts. Although A/B testing is becoming less expensive and easier to execute, the strategy still requires time and effort to implement, and the results of poorly designed A/B tests may be useless or misleading. Here are four tips to help marketers maximize A/B testing.

Think Small

To yield useful information, A/B tests have to focus on a single variable, such as an email subject line. If any other factor changes, such as the time the email was sent, you won’t be able to determine which factor produced a higher open rate. For this reason, A/B testing is best suited to small tweaks, like deciding where to place a call-to-action button on your web page, rather than major site redesign.

A/B testing websites

Focus on What Matters

With so much hype around A/B testing, it may be tempting to run a test to validate every decision you make. However, it’s probably a waste of time to A/B test a call to action button in two slightly different shades of purple. Focus your energy on tests that will provide the most valuable insights to your brand, such as what kinds of email marketing offers result in the most conversions.

Marketing team analyzing data

Pay Attention to Your Audience

Imagine you want to find out which promotional offer your customers are more likely to select. One offer is a percentage discount and the other offer is a free gift card. You run an A/B test and two out of three customers select the gift card. The gift card is clearly the winner, right?

You probably wouldn’t base a major decision on a survey of three customers. Don’t fall into the same trap with A/B tests. If you conduct a test among a small sample of customers, your results may be skewed and not hold true among a larger population.

This doesn’t mean that you need to have a ton of website traffic or huge emails lists to run A/B tests. However, you do need to make sure your results are statistically significant, as Ginny Soskey writes at Hubspot. Several calculators available online can help you determine whether test results are statistically significant, including this A/B Test Calculator from HubSpot. If your A/B test results aren’t statistically significant, think carefully about making changes based on them.

Business conference

Recognize the Limits of Your Test

Finally, it’s important to remember that an A/B test is a comparison between two options under a certain set of circumstances at one point in time. While A/B tests can provide helpful information, your test results won’t tell you why one option works better than another, whether customers will select the same option in the future, or if there’s an even better option than the two you are testing. When you recognize what A/B testing can and can’t tell your brand, you’ll be able to use of A/B test results wisely.

 Test limitations