Traffic splitters are used in AB testing when you want to try and make sure that the quality of the traffic you are sending to the landing pages you are testing is as similar as possible. The reason for wanting to do this is to avoid the impact of external factors on your test. So for example, you don’t want to test one landing page when the weather is horrible and your second one when the sun is out because such things, as many ecommerce businesses will agree, can have a surprisingly large impact on consumer behaviour.
Likewise, if you test one landing page just after people have been paid and a second one in the week before people have been paid, you are likely to see results that are substantially skewed simply because of the relative states of your customers’ bank balances.
Timing is not the only issue though: you need to make sure that the traffic you send to each page is demographically similar. It is no use testing one landing page that gets traffic from a rich part of town and then the second landing page with one that gets traffic from a poor part of town because the type of visitor will probably have a bigger impact than the copy you are testing.
To get round these issues, the best practice is to split traffic evenly between test landing pages on an alternating basis. Traffic splitters let you do this.
They receive incoming traffic and then display web page content that alternates, either by redirecting the traffic to one of two or more alternative URLs of by dynamically adjusting the content of the web page itself.
Traffic splitters are also known as URL rotators and you can try one out by clicking this link several times.