Not all digital marketers agree about the effectiveness of A/B testing. Certainly, when it’s done poorly A/B testing can be an unnecessary drain on time and resources; yet when A/B testing is done properly, it can have a significant positive impact on conversions.
If you’re unhappy with your current conversion rate, it may benefit you to spend some time thinking about improving your A/B testing approach.
What is an A/B test and can it really improve conversions?
An A/B test is a way to quickly compare two different versions of the same idea to see which one performs better. This kind of testing lets you focus on specific aspects of your site, app or content to find out where you have room for improvement.
When it’s done right, research has shown that split testing can boost conversions roughly 13% on average.
If you’ve been A/B testing for a while with limited success, you could be making one of the following mistakes:
- Not enough conversions to start
- Not testing long enough
- Testing the wrong things
- Using a sample size that’s too small
Doing A/B testing the right way
A/B testing is really only effective for sites with more than 1000 conversions per month. Anything less than that will yield statistically insignificant results.
You’ll also need to run your test long enough to reach at least 95% statistical significance. There are a number of tools you can use to measure this metric.
An experiment is only as good as its hypothesis, so if you’re not asking the right questions or targeting the right content, you won’t get very far.
Start with this simple formula:
- Identify your conversion problem
- Figure out how to test for solutions
- Anticipate the impact
The larger your A/B test sample, the better your data. Most experts recommend aiming for a sample size of at least 1000 subjects, though some suggest using as many as 5000. Remember that your total sample size will be split in half, so be sure you have sufficient numbers.
Research has shown that A/B tests can produce results when ran properly. Rather than waste your time and energy on a split test that doesn’t work, establish a solid plan that will help you ask the right questions, target the right data and get the results you want.