One of the most important things to do in marketing is split testing. In essence, this means that you create several different versions of your content and then watch closely to see which version converts the best. Why is this so important and how is it done? Read on to find out:
How Does This Work?
Split testing can take on any number of different kinds of methods. So for example, you might do a split test of your titles in the e-mails you send out. This is vitally important because you want to see how many people open the e-mail up with each kind of title. Once you get an idea of which one is getting more people to open up the e-mail, you know which version of your title you should use going forward.
You can also use split testing to check which version of a particular e-mail pitch is likely to convert better. In fact, it doesn’t have to be an even split either. You can do 2, 3 or even 4 different versions. In theory at least, you could dozens or hundreds of versions but practically, for most of us, 2-4 is more than sufficient to learn what works best.
Now, the way this works will depend on what you are using it for. If you are using split testing in order to see how many of your customers open up an e-mail you send out to them, you need to split your e-mail list into several sections. If for example you have a list with 2,000 names, you might take around 300 of them and then send out two different versions of your e-mail to 150 each. The idea is not to simply split your entire list because if you do that, you defeat the whole purpose. Your goal should be to see from a smaller test sample which pitch is likely to convert more often.
If you are trying to split test different versions of your landing page, you might want to get a special script which will randomly select one or the other version of your landing page to show to people who come to your site. Then, you need to track which one converted better (i.e. put people who sign up to your e-mail into two separate sublists so that you can see which one grew faster).
The Big Issue with This
The problem with split testing is that it’s easy to misinterpret the data – you may see a higher conversion rate with a particular version of an e-mail but you may not notice that you had a higher open rate with the other version of your title. You need to look closely at all the data you receive so that you can make an intelligent decision about how to move forward.
Of course, the other big issue with split testing is that there is always the factor of random bad luck involved. While it is extraordinarily unlikely when you decide to randomly run a split test, it is just vaguely possible (in theory) that one version went out to those who are mostly not buyers while another went to mostly buyers. However, the odds that this will happen when you do a random split are almost infinitesimally low so I wouldn’t worry too much about it.