If you've ever had your eyes examined, then you already know something about A/B Testing. Which do you like better? This one... or this one? A or B?
A/B Testing is also called "Split Testing" and "Bucket Testing." Some people prefer the term "Split Testing" because in fact, you can actually include more than two variations in a single test, or "experiment." For example, if you were to test four variations of a "Buy Now" button, it would be more like an "A/B/C/D" test.
Let's say that you want to test two different versions of something on your home page. Which is most likely to result in a customer making a purchase? Does the background color make any difference? What about the headlines and font size? The only way to find out is to try them and see what happens.
Although you can include multiple variations of a single item or page element in an A/B test, you are limited to testing one element, (such as a button) at a time in a single test. To combine multiple variations of multiple elements in a single test, you need a Multivariate Test.
Sometimes the results of an A/B Test will surprise you, because it isn't always a matter of design or aesthetics. Here are some reported results:
- Changing the Sign-Up button from green to red increased conversion by 34%.
- Displaying fewer options increased conversion by 20%.
- The words, "It's Free" next to a Sign-Up button increased conversion by 28%.
- Use of human photos doubled the conversion rate.
- A complete redesign of the product page increased sales by 20%.
How Does A/B Testing Work?
First of all, you start with the original page (or page element) as your "control."
Then you decide how to determine that the goal has been reached. For example, if you want to see how many people make a purchase, then the "Thank You" page can be the indicator that the conversion goal has been reached.
A simple test would present one version of the page to one visitor, and the another version to the next. Or it could swap out the test pages randomly, or be based on some other parameter or demographic. It all depends on the capabilities of the testing program and what you're trying to achieve.
How long should you run the test? For reliable results, you should be prepared to run the test for a month, or more. You can use this an online calculator to determine the recommended test duration, based on the number of visitors and expected conversion rate.
A/B Testing Tools
There are many testing tools on the market, but here are a couple of good ones that, as an added bonus, offer integration with Google Analytics.
Visual Website Optimizer is the tool of choice for many who want an easy-to-use, visual interface, combined with a rich feature set. It's available by monthly subscription, with a thirty-day free trial.