If one of your website pages isn’t performing as well - bounces are high and conversions are low - you may turn to A/B testing to determine the issue and improve the user experience.
The process of A/B testing involves showing two versions of your web page, version A and version B, to your audience with minor changes in order to analyse which performs best.
It is essentially experimentation with your own web pages. Let’s go back in time to GCSE science...
Before you start any experiment, you must have a hypothesis. It’s important that you don’t blindly run into A/B testing and that you have logic and data to back up the changes you wish to test. Dive into your website’s analytics. Are there any blatant issues? Does one page outperform the others? Dig in deep to try and understand what elements are affecting performance.
At this stage, you may also want to ask your users for feedback with a small survey to collect useful qualitative data. It can take some guessing work out of your hypotheses as they may be able to identify issues that you weren’t aware of.
In order to run an accurate A/B test, you must have a control web page (one that you shouldn’t touch) as well as a variation web page (with a minor change). When a user lands on the page, they will be shown either version A or version B chosen at random and their behaviour will be analysed.
Running frequent A/B tests will enable you to optimise your web pages based on visitors’ behaviours and actions, ultimately improving the user experience.
But what can you actually test? It’s important that you make one small tweak rather than multiple changes as this will allow you to determine which variation your audience prefers.
To get you started, we’ve come up with a list of web page elements that you might want to consider for your A/B test...
Headline
Your headline is the first thing that a visitor sees, appearing ‘above the fold’ on your web page. For this reason, it is paramount that it intrigues your visitor as well as demonstrating what information the rest of the page includes. Ask yourself - does your headline make an impact? Do you have a particularly high bounce rate for a certain page? A/B test your headline to encourage your visitors to read on.
CTA
A good call to action should be explicit and clear as to what the visitor is clicking to do. You may want to experiment with different verbs to see what people respond to best; ie. ‘read more’, ‘discover more’, ‘download your resource’. You can also create a sense of urgency by adding ‘now’ to your CTA, encouraging your visitor to act fast or risk missing out.
Images
They say “A picture is worth a thousand words” and a poor quality image can negatively impact a web page. While buying stock photography can save your business time, it may affect the look and feel of your brand - is it genuine? It is important to analyse how your audience responds to your images as they should be intricately placed to support your text, not distract from the main call to action for each individual web page.
Copy Length
Sometimes it’s hard to tell if you have conveyed your products’ or services’ message to your visitors the right way. Too long-winded? Too vague? Find your sweet spot by A/B testing your copy length.
Number of Form Fields
Not driving enough conversions on one of your landing pages? Your form fields may have something to do with this as your offer must be as valuable to the visitor as the information you seek to get in return. If you think you may be asking for too much, try experimenting with removing some fields to see if conversions start to rise.
Alternatively, if there is a piece of information that would be vital to learning more about your leads/customers, you may add in a form field to see if conversions are impacted.
Once you have decided what element of your web pages you want to A/B test, you may be thinking, “What is the minimum amount of time I should run my experiment for?”. The easy answer is there isn’t! Depending on the amount of website traffic you receive, the test could last anything from a week to a month. When you have enough data to determine a winning variation, you can stop the experiment and publish the best performing page for all to see.
