Everything you need to know about A/B testing

As the goal of any business is to thrive and see your revenues reach new heights, it is essential to constantly perfect your company website in order to realize this ambition. To do this, it is necessary to test your website using A/B tests. But what is this exactly?

What is an A/B test?

In general, an A/B test consists of testing several versions of the same website or webpage to identify the one that will generate the best conversion rate, which is the increase in contracts, sales or rate of enrollment.

What should be tested?

To get a meaningful comparison report, target elements that are likely to have a real effect on your audience. For example, it is unnecessary to include the size of the font or the working of a paragraph whose information is of secondary importance. Your focus could instead be on the visual appeal of your web page as well as on the central part of the presentation of your services or products, or on some of your tabs.

The various tools at your fingertips to carry out your A/B test

Google Optimizer

Google Optimizer will allow you to compare a large number of versions of your website and will indicate plenty of information (percentages) divided into 4 sections.

• The first column represents the conversion rate of your original version and subsequent versions;
• The second column evaluates the likelihood that the modified version will perform better than the original version;
• The third column indicates the chance that a modified version will perform better than all the other versions contained in the analysis;
• Finally, the last indicates the improvement of the conversion compared to the original version.

If you decide to use Google Site Optimizer to perform your A/B test, you should be aware that your HTML files must be uploaded (uploaded) to your server. Nevertheless, the advantage of this tool is that it is free.

Looking for a marketing expert who can conduct A/B testing on your website?

Contact us so we can refer you to specialists in our network

Optimizely

More advanced than the Google Optimizer, Optimizely allows you to easily make changes to your page without having to edit your code. In addition, you will be able to follow several conversion goals and test one aspect on each page of your site simultaneously. Unlike its' predecessor, however, you will have to pay when following the completion of the trial period. It is worth noting that Optimizely installs cookies onto the browser of visitors who arrive on your website.

General technical aspects

For each of the tests you perform, a percentage representing the probability of reliability will offer you the odds that this version will surpass the original version. If you receive multiple transformations on a version (registrations, sales, or concrete actions from the user), the efficiency of this version will be significantly higher than that of the opposing version. 

The question now is: “How long do we need to collect data?” It is recommended that you continue picking until a 95% reliability rate is achieved.

What if your A/B tests record very few transformations on your website?

In order to avoid waiting an eternity before obtaining results that are statistically interesting, you can make your comparison based on more subtle elements. This will make it possible to identify precisely which aspects give rise to a difference in behaviour among the users.

The steps of an A/B test

The first step is to collect data that allows you to identify the elements that require improvement, such as an analytics tool like Google Analytics. The latter effectively highlights the different behaviours displayed by visitors to your site, making it possible to target the weaker pages. Google Analytics is free, and it is therefore worth trying out.

As with any project, you will have time to set goals for the conversion rate generated by the modification of an element of your website. In order to obtain useful and relevant information, be sure to set goals that are sufficiently precise and concrete. 

Next, it is time to generate ideas about new versions of your page or website, and these will be tested and compared with the original version. Make a list and put the most effective versions at the top.

Now, use your test software (Google Site Optimizer or Optimizing) to make the changes mentioned, building the versions that you want to test. That decisive moment to launch the versions you’ve created has arrived. 

Visitors who arrive on your website will be directed to its original version or to its modified version, though chance will determine which of the two will be consulted.

The behaviour of your users will determine the conversion rate generated by the highlighted targeted elements and compile them. Now that results have been collected, you can analyze them. It will be clear to identify the performative differences between each version and find your winner.

Some additional information about A/B testing

A short note on the conversion tunnel

Although the data that we are interested in is often the conversion rate, it may also be relevant to analyze your conversion tunnel. This is the set of steps that are travelled between the arrival of a user on your site and a concrete gesture made by that user.

Pages that have little traffic

Alternatively, remember that pages that have little traffic will only provide you with little information about conversion rates as well as user behaviour. To maximize your time, therefore target the pages of your website that are the most popular. 

If you still want to test your less popular pages, do not offer more than three tests, as this will result in splitting your traffic too much. This would greatly slow down the collection of data.