Who doesn’t want to increase sales on their online store? A great start is to find ways of converting more of your sites visitors into customers. This process is known as conversion rate optimisation (or CRO) and right now there’s a whole bunch of services online that can be useful in trying to achieve this.

These services can provide insights and tools that are invaluable when making decisions related to your store. Inevitably though, these services are often presented as a sure-fire way to increase sales. Simply alter this line of text or change that button’s colour and watch the money roll in! Of course though, it’s not that simple.

An important methodology you’ll undoubtedly come across when reading about conversion rate optimising is AB testing (or split testing). This method is one of many, but it will be the focus of this article. I’ll explore the basic premise of AB testing, things to keep in mind when performing tests and also some online tools that will allow you to get the insights you need to turn your eCommerce store into a sales beast.

AB testing is a very simple concept. It is an experiment with two variations. Typically one being the control and the other being the treatment. An example of an AB test on an eCommerce site would be half of our visitors with the existing site and the other half with a modified version of the site. While doing so, we would measure which version gets more customers through the checkout, to the receipt page.

In theory it can be a great way to gain insights into the behaviour of site visitors. In reality, the results you find can be misleading or inconclusive. Also, changes you make to the treatment or B variation could end up costing you money instead of making you money.

Here some important things to keep in mind if you plan to employ AB testing on your eCommerce store:

  • Know your baseline. Analyse where the store is at already. Identify what you believe it’s strengths and weaknesses are. Have data and analytics to refer to.
  • Split A and B evenly. With an even split half and half, you can be sure you have the largest data set for both tests.
  • Test A and B at the same time. Too many variables will come into play if you test alternating hours/days/weeks.
  • Set a goal. Know what you are trying to achieve.
  • Make a hypothesis. Predict what you think will happen. Not just what you want to happen, but what could go wrong and what other factors may influence your test.
  • Time is required. There is no quick test that will provide adequate insight.
  • Focus on big impact. You can test anything, but small tweaks can wait. Prioritise targeting those big strengths and weaknesses that have been identified.
  • Data can be misleading. Sometimes your results can reflect your hypothesis but for completely different reasons. There are too many variables for things to be predictable.

And most importantly, your AB test is only as good as your B. Meaning if you haven’t put enough research and reasoning into your B variant, it could cause you to start losing sales rather than attracting them.

Finally, let’s have a look at some of the great tools available to help you starts performing AB tests on your store:

  • Visual Website Optimizer

    A feature rich cloud based subscription service that is a complete solution for AB testing. Simple add the provided code to the header of your site and start testing.

    • Great variety of analysis and reporting tools
    • Non-developer friendly
    • Great visual editor for easily making variations
    • Clickmaps and heatmaps
    • WordPress and Magento integration
    • Guides you by making you set goals
    • Adds further javascript to your site
    • No free option
  • Optimizely

    Another cloud based service that provides a variety of tools. The most important difference being they offer a free option. Also as easy as adding the provided code to your site and beginning experimentation.

    • Visual editor for easily making variations
    • Detailed reporting page
    • Non-developer friendly
    • Visual editor for easily making variations
    • No clickmaps or heatmaps included
    • WordPress and Magento integration
    • Adds further javascript to your site
    • Free starter option with some limitations
    • Integrations panel for integrating with other services
  • Google content experiments

    The home for internet data analysis. Plug in the Google Analytics code into your site and go to the “Content Experiments” feature

    • Completely free
    • Goal setting features
    • No visual editor
    • Provides raw data
    • Not friendly for non-developers
    • Lacks integration with other data services
    • Makes the website URL look messy by appending the parameters to it

In summing up:

There are services out there for both developers and non-developers for split testing. Anyone comfortable with web development who wants to do some testing is probably best to use Google content experiments rather than investing in a paid service. Sites should already be hooked up to Google Analytics, so you’re already halfway to starting an experiment. VWO and Optimizely are very valuable with their simple interfaces and quick setup. They are incredibly useful for someone who doesn’t have web development skills and offer many other attractive tools.

In the end though, the value of these services is really dependant on how much research and work you are willing to put into your tests. But when done correctly, they can provide you with great insight and lead more of your customers to the checkout.