For improving the conversion rates of specific pages or your business site as a whole, it is vital to choose the right type and mode of experiment, at the outset. We shall consider a few key aspects to grasp what test is suited for your purpose so as to take an informed decision before you go ahead with testing.
At the outset, we need to realize that both A/B testing and multivariate testing are suited to different kinds of conversion parameters, relevant experiments and sites.
A/B tests let you analyze two entirely different variations or versions of a particular page from the conversion point of view. You might need to opt for this mode of testing in case you are looking to move certain things around on a webpage or perhaps try out a different layout.
You can consider A/B tests in case you wish to alter the look and feel of a webpage. You may also use them for creating and publishing newer version/s of any existing page. On the other hand, there are different scenarios for which you might want to employ multivariate tests. These tests allow you to try out different possible combinations of the various content types on the very same webpage.
You might look to go for a multivariate test in case you are getting a rather high volume of online page traffic (say, more than a thousand pageviews a week). You might consider it if your page layout and its overall design is going to remain the same. Another scenario for a multivariate test is if you wish to change certain specific areas or elements of your webpage (such as an image or a headline) simultaneously.
So, traffic is a factor that will decide if you select an A/B test or you go for a multivariate test apart from the aspects to test. In case your website generally has a low traffic flow and you are looking for quick results, former is an advisable option to analyze as well as compare different versions of your advertising content for its effectiveness in terms of generating quality leads and referring customers.
To start with, simply change a few elements between different versions prior to launching an A/B test, for instance a long versus short form, or a shorter title versus a longer one. You may alter a few aspects in a layout to test them against another by keeping it static. Ideally, run your test for a week or so, at least. You may conduct it until certain amount of conversions take place, which are good enough to draw clear assumptions so as to analyze the user flow and study the results. We shall get more ideas to select the right type of conversion tracking experiment in the next post.