By Yongzhen Tham on Mar 30, 2015 9:30:00 AM
A/B testing is a form of statistical hypothesis testing, which is carried out with the goal of identifying what changes you should make to your website in order to optimize it for maximum interest from its viewers. If you haven’t heard about this concept before, here are 6 things to know about.
1. The control and the variation
The control is simply your “A”, the version of call-to-action (CTA)/landing page/form format etc. that is normally being used, and the variation is the “B”, the version of changes you are trying to test.
However, one should always remember that just because it is called A/B testing doesn’t mean you can only test one variation against one control; if you’ve previously tested a blue background against a grey background with the blue background proving to generate more conversions, it does not mean that blue is your final answer. Instead, you could then go a step further and test your blue background against other colors, in order to find the most effective color for increasing conversions.
2. Testing should always be conducted simultaneously
In A/B testing, there is always the control and the variation that is being tested for its respective effectiveness. Don’t test the variation after having tested the control, or vice versa, because there are always other factors besides the testing that can affect your click-through rate. Test them at the same time.
3. If you have more than one difference, your results cannot be attributed to only one difference
For example, if you’ve decided to conduct your testing with two completely different pages, your results cannot be attributed solely to the CTA variation. Instead, you must consider that your results can be due to either the new CTA, form format, images, or background color, or a combination of all.
4. One A/B test is not enough
The process of testing should be a continuous one. Trends are always changing and people’s tastes are always changing. With that, the effectiveness of things are always changing as well. Carry out testing regularly in order to keep your website consistently at the top of its game.
5. You need to control the duration of testing
If you run your test for too short a period, your results are not going to be accurate. For example, if you conduct your test from Monday to Wednesday, you are missing out on results from people who usually browse the Internet on weekends. That being said, it’s also not advisable to run it for too long a period of time, because that means there are going to be more variables that you’re not in control of. To tackle this, make sure you’re aware (to the best of your ability) of any secondary variables that could be affecting your testing so that you’re able to account for any anomalies in your results.
6. A/B testing results are not an absolute reflection of everything
If you test a red CTA button against a green CTA button, with the red leading to more conversions, it does not mean that having red buttons and red text all over your web page is guaranteed to generate more conversions. While your test has shown that more people prefer the red to the green, there are still some who don’t. So don’t take one A/B test result and assume it is applicable to everything on your web page...it isn’t. Instead, carry out separate tests for different things, as some people may like a red CTA button but prefer a blue background on your landing page (read more about CTA A/B testing at "6 CTA Tests for Maximizing Click-through Rates").