A/B testing is a process where 2 variations of an element are tested against each other over a chosen period of time to determine which variation of the element yields better results.

You create a second homepage for your funnel and make your desired changes there. Once you publish, visitors are automatically directed to one of the two homepages.

Over time, in the analysis section, you'll see the variation that leads to a higher conversion rate.

This process allows you to continually test new hypotheses optimizing your funnel's performance.

An example:

Imagine that your funnel's conversion rate is 2%. Your goal is to significantly increase this percentage but you're unsure of the changes needed to achieve your desired increase.

As a way to find out, you set up an A/B test and modify the value proposition on the variant. Now you have created two different value propositions to test against each other.

Testing this over a 2-week period with 200 visitors to the original and modified funnel, you'll see that your second value proposition was stronger as more people visited that funnel and the conversion rate for that page increased from 2% to 6%.

You then end the test by declaring your variant as the winner. From now on, you choose to work with the version that led to the increased conversion rate.

Once you have created a modified version of your funnel and published it, head over to the analysis section of your funnel. Here, you can not only see the total page views and conversion rate of your entire funnel, but also access two evaluations - one for the original funnel and one for the variant.

This way, you get a chance to quickly identify the difference that each version has on your total page views and the overall conversion rate. Once you are satisfied with the results and decide on a clear winner, you can end the test and keep the winning version moving forward.

It's important not to modify the variant too much from the original so that you obtain a meaningful test result. For instance, simply changing the button color and testing which button color performs better is already sufficient. If you make significant changes to the variant, such as replacing multiple blocks, changing content, color scheme, etc. at the same time, you might lose your ability to determine which changes led to the results when evaluating the test.

Find additional information on the following topics in our Help Center: