• Share
  • Share
  • Share
  • Share
  • Share

Posted by Melanie Smith on 24 June 2015

We take a look at different website areas that can be tested & highlight some successful examples

Website A/B testing sounds like quite a technical testing method, but actually it’s a fairly straightforward way of gaining valuable insight into how your website visitors behave, and most importantly is an effective way of increasing conversion rates.  

In simple terms A/B testing (sometimes also called split testing) involves monitoring two or more versions of a website page element or design over time and assessing which performs better.  The assessment is judged against a metric that defines success - such as website visitors clicking a Call to Action (CTA) button. The idea is that you test different versions of your website at the same time in order to see which variation has the most success. 

Example of how two layouts may vary for A/B testing

How can A/B testing help?

A/B testing is useful for monitoring user activity and establishing if a design change has improved the customer experience or had a negative affect. This is especially important for e-commerce businesses as a new element introduced to a website has the potential to disrupt a user’s shopping experience and have a negative effect on sales. 

In this post I'm going to take a look at some different types of A/B testing and highlight some real life examples of the benefits. 

Call to action (CTA) text changes

A simple text change of a CTA button can drive significant changes in user behaviour.  Making a button less or more encouraging can majorly impact on how a user interacts with your website and improve conversion rates.  Therefore assumptions should not be taken that you have the most effective wording for your CTA buttons, test alternatives to see which brings the best results.     

ExampleSchuh, the footwear retailer used A/B testing to discover that when they made the simple change of altering the wording of their product page CTA from ‘Buy now’ to ‘Add to bag’ that this led to a 17% increase in basket additions.  

Call to action (CTA) design changes

Even a small design change such as altering the colour of buttons can make big differences so it is worth testing to see what success you get with different design options

Example: Testing conducted by Hubspot identified that changing the colour of a CTA button from green to red increased clicks by 21%, therefore improving the performance of the web page with a very small change. The two pages tested by Hubspot are shown below:

Example of A/B testing with different coloured buttons

Headings and banners

As the first thing a user should see on your website, headings should be clean, easy to digest and eye catching, but you can’t always assume that just because a banner is in a prominent position on your website it will make your visitors behave in a desired way. 

Example: When video games developers EA Games launched SimCity they used a banner advert to drive pre-orders but were disappointed with the level of response they received. They decided to A/B test the banner with several variations such as changing the placement, colour and display of the offer, and even went to the extent of removing the promotional offer from the page altogether. The results of their testing surprised them as they discovered that removing the banner totally actually drove 43.4% more purchases, suggesting that the banner was actually not required at all and visitors felt it added extra clutter to the page.

Form layout and design

The user experience of forms should be considered carefully when making any changes to a form layout or design and are a prime example of where A/B testing is valuable to check that changes made haven’t had a negative impact on visitor experience. 

Example:  After examining analytics data Paypoint identified that their checkout process was leaking sales. They recognised that their biggest opportunity was to rework the data collection page so they spent time focussing on the flow of the form and A/B tested variations of the ‘your details’ page.  The winning variation increased the conversion rate of PayPoint’s online application process by a staggering 159%.

Content style changes 

Different variations of content or text style can also be tested using A/B testing.  Checking how long a user stays on a website page can be a good indication that there is an issue with the content or text style - if they find content hard to read they will quickly leave and find information elsewhere.

Example: Underwater Audio updated their information section to convey a clearer message as they felt their product comparison content wasn't engaging enough. They wanted to test whether more concise information about the products would enable customers to make a decision between products more easily. Changes they tested included moving all of the information into the table and not in paragraphs underneath, dividing the table into different specific categories, making the product features and specifications more detailed, and instead of showing several call to action links on the page they tested having only one link to each product page.  They also cleaned the table layout and changed the font of the page to be consistent with the rest of the site.  The result was that the more focused page with improved copy and design increased website sales by 40.81%.  The different options tested are shown below:

Example of an A/B test on different table styling

This area of testing involves examining how conversion results are affected by altering an element that is used on multiple parts of a web page. 

Example: Who Accepts Amex is an affiliate model website which lists companies who do and don’t accept American Express.  Their revenue comes from commission earned from sales on other websites. Therefore, any opportunity to maximise clicks on their website is hugely important.  Through data analysis they discovered that the ‘companies’ page template of their website produced fewer external clicks per page compared to the ‘categories’ template despite having more page-views.  This prompted them to perform optimisation testing on this page.  They decided to test the impact which changing font size throughout the page had on clicks. The goal of this test was to track the number of external website clicks. The control, which had website links in font-size 14px, was tested against 6 variations from font-size 12px to 18px (with 1px increments).  After 28 days, they had a clear winner of the variation with font-size 18px recording 32.68% more clicks.

Tools to help with A/B testing

To test your website you first need to decide which aspect(s) you are going to change, and then each change must be tested with an equal proportion of your website visitors. 

Next a goal should be set that you wish the user to complete, such as clicking through to the success page after making an order.  This lets you assess how successful the conversion rate is for the variable compared to the control version. 

There are many online tools that can help with A/B testing, displaying results and making the data easier to understand. Here are a few examples:

  • Google Analytics – a great free service that provides comprehensive analytics and allows you to access Google's data pool. Within Google analytics there is a really useful feature called 'Content Experiments' which allows you to simply create test variants of a web page and visit your results.
  • Kissmetrics – lets you monitor data and is very easy to set up on a web page.  You simply copy and paste a piece of JavaScript onto your webpage which allows you to see and compare a huge amount of customisable data in easily digested graphs. Although it is a very nice tool it does come with a monthly cost. 
  • Optimizely – another easy to set up tool that only requires you to insert a piece of code into the page you want to test. Optimizely allows you to customise what you wish to track and set timelines for your tests.  It also has a nice feature that allows you to visually edit your control version without any code, so a good one to consider if you aren't a developer.

If you are interested in having a chat on how the Status team can help you with A/B testing your website in order to improve conversion results then please get in touch

Further Reading

And if you would like to explore this topic further why not take a look at the following:

Comments:

smartly dressed man holding his mobile phone in the air and shouting at it

How to improve website speed

Find out why website speed matters - and the steps you can take to improve your site

Featured Post Reading Time

vertical stack of various different newspapers

Why we should say goodbye to the fold

With so many different types of device screen sizes to cater for we look at why there's now no such thing as 'the fold'

3 minutes

Two chess pieces face off in competition

Hubspot COS Vs a traditional CMS

We compare the features of inbound marketing tool Hubspot COS to a traditional CMS

10 minutes

Interested?

If you think we might be a good fit for your business, fill out the form at the link below and we'll send you our agency playbook. It contains lots more information about us and what we're like to work with.

Send me the playbook