If you could have one superpower, would it be mind reading? Humans are unpredictable by nature and their behavior is far from consistent. This comes as a big problem to digital marketers and online businesses who just want to get through to as many consumers as possible. Well, Google realized this problem nearly 17 years ago, when the company ran its first A/B test on February 27, 2000. So, if A/B testing has 17 years of successful experience for Google, why isn’t it used more often for digital marketing campaigns, websites, and ads?
What is A/B Testing?
A/B testing is a comparison of two versions of the same website, newsletter, ad, etc. One version is considered the control, which is the website as it currently appears, and the second version is a modification of the website, which includes variation in content from the original site. The original site is seen by one group of viewers, chosen at random in order to minimize statistical error, and the modified site is seen by an identical group. Viewers are not informed that they are looking at a varied version of the site. The version that most effectively achieves the site’s set objective is deemed the “winner”, telling the web designer which features should be changed in order to optimize conversions.
Where do I start?
When you running an A/B test, you are running an experiment. Think back to elementary school when you first began experimenting in science and the scientific method. That’s right, A/B testing is not difficult — all it requires is some strategy and going back to the basics! Remember, like any other experiment, it is best to change one variable at a time.
- Think of the purpose of the A/B test. What is the problem with your website? And, more importantly, what is the result you hope to achieve by employing A/B tests?
- Research. Use analytics to find out what features need modification. The most impactful areas to consider are pages that have high drop-off rates (see our blog about Heat Maps), pages with most the most call-to-action buttons, and areas most likely to generate data.
- Predict a hypothesis for the expected change in customer behavior.
- Experiment with the original site and modified site. It is important to allow enough time for the test; however, too much time allows external variables to influence the results.
- At the conclusion of the test, analyze results to decide if the modification led to the desired outcome.
- Conclusion. Not every test will be a success, and that’s okay. If a test results in fewer clicks, subscribers, or any other objective, or is not statistically significant at all, then that can point you in the right direction for the next A/B test. Multiple tests can be used to pinpoint the most effective aspects of your site, so you can combine these to create the optimal user-friendly site for your product or service!
What Variables Can I test?
- Headings and Subtitles
- Product Descriptions
- Copy
- Offers/Deals
- Prices
- Images/Graphics
- Page Layout
- Color
- Call to Actions and Buttons
- Element Location
Results
Results will vary for every website, campaign, and ad because no customer profile is made up of the exact same demographic. For this reason, it is difficult to share viable statistical results.
With A/B testing, marketers and web designers are able to make mindful changes to websites and analyze that data to optimize customer interaction. It is a proven and easy way to learn what impacts behavior most. Blind guessing can produce more harm than good, so contact 10 Pound Gorilla to help you optimize your website, email campaigns, and advertisements!