A/B Testing Explained

If you want to maximise sales from your website it is important that you fully understand A/B Testing. It’s pretty simple and a great way to find out what works well on your site and what doesn’t. It gives you key information about customer behaviours so you can fine tune content and increase your conversion rate. A/B Testing (also known as Split Testing) shows that sometimes, even making a small change can make a significant improvement.


Contrast and compare

It works like this; for your site you simply create two versions of a page or a function within a page. On your site you may have a button to click for more information or to make an enquiry. A straightforward test would be to create a page with the button or fonts in red and another in green. A/B Testing will monitor use of both and then tell you which is most used. It allows you to test something as small as a navigation button (which may be a significant contributor to your site’s return on investment), or layout, colours, copywriting, graphics, photographs, a whole page or even an entire site. Key metrics such as conversion rate, sales, bounce rate, etc., can be reported, allowing you to then install the best performing version. A variety of software is available to support an A/B test. Visual Website Optimizer and Google Website Optimizer are both free and good places to start.

Let’s get testing

To test (using our simple example of an enquiry button) in html, you create your alternative version or versions of your button. The software will change the button seen on the page over a period of time. It will test both versions simultaneously and then report on use. If you want to test an entire page, you create two URL versions and your A/B tool will direct some visitors to one, an equal number to your alternative, and then report back on activity.

How long should I test for?

Think carefully about the time period for each test. You should base it on the comparison you need to know, the existing levels of site traffic (so you can ensure you will have a reasonable minimum use to enable you to see a comparison) and the overall time period you’ve allowed for conversion testing. Overall, ensure you can produce quantifiable reporting to give what is called ‘statistical confidence’. Use acalculator to determine exactly how long to run a test.


Visitors and variations

If you are testing a core part of your website, your site visitors may see a variety of page versions. This could be confusing and off-putting. However, you can set up A/B Testing so that only new visitors are included. Your tool should also have a mechanism for remembering which variation a visitor has seen and showing them the same one next time they visit. We tested FCL’s Freight Quote form form rigorously and found a notable increase in conversions when it had no navigation.


Be creative

A/B Testing gives you an ideal opportunity to experiment and try out a variety of different site functions. For example, FCL Global if you want customers to make an online enquiry, would they be more likely to click a button saying ’email our online expert Katie now’ (with of course, a smiley photograph of Katie) rather than simply, ‘contact us?’


Keep testing; you’re bound to find improvements

A/B Testing is a key tool to help you measure and improve your optimisation and conversion rate, and that means a boost to your sales or other website goals. To give you an idea of what other people are doing, a number of useful A/B test examples are linked here.


Writing Decisions: Headline Tests on the Highrise Sign-Up Page

Human Photos Double Conversion Rates

Two Magical Words Increased Conversion Rate by 28%

Changing the Sign-Up Button from Green to Red

“Mad Libs” style form increases conversion 25-40%

Complete redesign of product page increased sales by 20%