Split Testing is a must-know website optimization technique that will get you more conversions. Read this, and in 5 minutes you'll be a pro.


What is split testing?
Split testing serves different 'variations' of your website to visitors, and lets you track how successful each variation is. Think of split testing as a controlled experiment, where your users will provide all the data for you. Free software will take care of all the technical work; we'll get to that in a bit. First, let's cover the basics.

There are 2 different flavors of split testing. Each has their own benefits and drawbacks. Here they are in a nutshell


A/B Testing
A/B testing is the vanilla of split testing flavors. It tastes OK, it does the job, but it won't knock your socks off with large amounts of data (which can be a good thing).

Use this method when you don't have a lot of visitors coming in, or if you only want to test 2 different variations of your page at a time. Let's use an extremely basic example to demonstrate. Suppose you have 2 different logos, and you want to see which one leads to more conversions. You would define each 'variation' of your page within the split testing software, then you would serve each of the variations equally (each variation would be shown to 50% of visitors). After a few hundred (or thousand) visitors have used your site, you will be able to compare which logo was more effective. You can disregard the poorer performing logo, and work off the better one.

A/B testing can be used to test logos, headlines, closing sales statements, offers, etc. Pretty much anything you can think of can be tested. The results show up within the free software. This is the simplest form of split testing.

The big thing to know with split testing is that you can only test different variations of a webpage. Multivariate testing is the other technique we will examine, and it allows you to test different variations at an elemental level.

To illustrate this difference, we could test 2 different pages. On page 1, we have a blue logo, with a very comedic headline. On page 2, we have a black logo, with a very serious headline. Let's suppose that page 2 outperformed page 1. With A/B testing, we wouldn't know exactly why page 2 outperformed page 1: was it because of the black logo, or the serious headline? Or both? This is where multivariate testing comes in.


Multivariate Testing
If A/B testing is vanilla, multivariate would be an organic fruit sherbet. It would take a lot more work, but it will provide you with a more complex dessert; perhaps something for those with finer palettes.

Multivariate testing is only really practical if you have a lot of visitors coming into your website. It basically extends split A/B testing to allow for more than 2 different variations at an elemental level. The testing software will then serve different versions of your page, with each element being 'scrambled'. Here's a simple illustration:

MyTestWebpage

Headline (1 of 3 will be served)
This is my first test headline
This is my second headline
This is my third headline

Closing Statement (1 of 2 will be served)
This is my first closing statement
This is my second closing statement

Price (1 of 3 will be served)
This is my first price
This is my second price
This is my third price

This MyTestWebpage shown above has 18 different combinations to be served up! You could use headline 1, with closing statement 2, with price 3; or 17 other combinations! Conversely, A/B testing will only have 2 at a time. This is important because to get any meaningful results, you'll want to see how each variation performs at least a hundred times. With A/B testing, that means you'll need about 200 visitors. With multivariate testing (for the MyTestWebpage example above), you'd need about 1700 visitors! A very big difference.


Free Software you'll need
Google provides free split testing software called Google Optimizer; it's easy to install into your site, and it's very easy to use. Installation and setup instructions can be found here.


Setting up an A/B experiment video
Here is a video that displays exactly how to set up an A/B experiment:

Setting up a multivariate experiment
Here is a video that shows how to set up a multivariate experiment:

Please press Tweet if you enjoyed this!  Tweet

Prefer Facebook? Click to Like this!