Guest article by Mark Baartse of TheHolisticWeb

Doing split tests on your online shop is one of the best ways to increase your profitability. If you’re not familiar with split testing, there’s two basic types of tests.

A/B test – this is when you make two different versions of a page. Half the visitors randomly see one version, and half see the other version. The two versions have differences such as different layouts, colors, copy, etc. You set a goal for the test, what it is you want to improve. Usually this goal is pretty obvious – completing a sale! You then run the test. The software sends half the people to one version, half the the other, and measures which half buy more. Often small changes can make huge differences in sales.

The other type of test is a multivariate test. This is when you have a few “zones” on one page and test different content in those zones. For example, you might have an Add to Cart button zone and a Headline zone. In the Add to Cart zone, you are testing 3 different “Add to Cart” buttons and in the headline zone you are testing two different headlines. The software randomly rotates all combinations of the different content in each zone and tells you which combination results in the most sales.

The goal of testing is to improve your conversion rate. Most shops are converting around 2% of their visitors. If you can increase this to just 2.1%, your revenue has gone up 5% – without any increase in marketing spend! It’s not hard to increase your conversion rate 5%, 10%, even 20% by running a few well planned tests. The testing professionals sometimes see improvements of over 100%!

While there’s lots of software to run tests with, the most popular is Google Website Optimizer which is powerful and free.

Here’s where it gets a bit tricky. Most web pages which talk about testing use landings pages as their example – a single page with a single purpose. These pages are easy to test. You create two versions, and you’re done. In an online shop, you often get more value by testing templates, such as your product page template. But how do you do this?

It’s not as hard as it sounds. It will require some basic technical skills, but you don’t need to be a hard core programmer. If you are scared of code, a programmer should be able to help you set it up in no more than an hour or two.

Let’s say you want to test a different “Add to Cart” button – a green one and a blue one. Creating a second template and running an A/B test is very difficult in most shopping carts. The simpler alternative is to use a multivariate test. Since you are just testing one section of the page, then it’s in practice much like an A/B test. However, in a template, a multivariate test is technically much simpler.

All you need to do is sign up for Google Website Optimizer. Create a new multivariate test. You’ll need to go through the various steps (Google Website Optimizer has some good help files if you get stuck). Once you have the code that Google Website Optimizer provides, you need to insert the code into your actual template file. While Google Website Optimizer looks like it is geared up for just one page, it works totally fine when the same code appears in a template which can make up 100s or 1000s of pages, no problem. Put the goal completion script in your checkout’s “thank you” page, and then you’re running!

Typically you’ll need to wait a few weeks or more till you get meaningful results depending on how drastic the change is. If the test is only improving things by 1%, it might take months before you have an answer as to which is the better version, but a change of 10% could be a week or two, depending how many sales you have.

An alternative is a few shopping carts such as BigCommerce now have support for Google Website Optimizer without having to mess around. This really encourages testing and should help you improve your conversion rate even faster.

I strongly encourage you to experiment with split testing as a great and cheap way to improve your store’s conversion rate. If you aren’t sure what to test, a good book is the aptly name “Always Be Testing” which has some great examples of tests to run.