Thursday, October 15, 2015

A/B Testing Has Arrived

Cambuilder A/B Testing

What is it?

Cambuilder A/B testing compares two versions of a site to see which one performs better in joinsnew spenders, andrevenue. Visitors to your site are divided equally into two groups. One group views version A, the other group views version B. Cambuilder gathers data so you can monitor the progress and make an informed decision about which design to use.
*Tip: Changing only one design element at a time makes it easier to track your data:
A/B testing example 1

How long do I run the test?

The suggested running time for this type of A/B test is between 7-30 days depending on the volume of your traffic. Any less might make it hard for you to draw a reliable conclusion. It is acceptable to run the test longer, however, you may want to limit the timeline to 60 days for best results.

How do I use it?

1. Choose an approved site from the white label editor by clicking "Edit."
A/B testing example 2
2. Make desired changes to your site by using the site editor.
A/B testing example 3
3. Click "Publish"
A/B testing example 4
4. Select "Start A/B Test" button from the pop-up window. This will create a new site — “B” — using the changes you just implemented. It will also keep your original site — "A." If you decide to cancel, your original site will simply be reinstated. Clicking the "Publish" button will apply your recent changes to your current site and will not create an A/B Test.
A/B testing example 5
5. Once your site is approved, click the "View test" button via the white label editor to enter the A/B test environment.
A/B testing example 6
6. In the A/B test environment watch the progress between the two site designs. When you’ve made your choice, pick the winning site by clicking "Use site." Stop the test at any time by clicking the "Stop test," button. Stopping a test will revert back to your original site — "A."
A/B testing example 7

What to expect

During A/B testing, one skin may appear to perform better than the other. However, because of the nature of statistical data, these changes may have little or no effect on results. Although Cambuilder cannot recommend choosing a specific site design, we suggest you make your own choices based on the results presented.
If the results do not give a clear indication of a winning design, we recommend you either continue the test for a little longer or cancel it and try doing another test with different changes.