A/B testing: an essential tool for eCommerce
29.05.2016
ab testing preview

In order to develop a web‐project, it is necessary to constantly work on its improvement by conducting various marketing and user experience researches. There is wide selection of tools and services that can make these improvements efficient and time-saving. The most effective applied method of improving the quality of a website is A/B testing. It is also called split testing. This is a marketing method for evaluation of the effectiveness of web pages and elements. According to the obtained data you can implement those components that have a positive impact on the result, whether it is a conversion, the average check, the visit time of the users on the page, bounce rate, and other metrics. And it can also help you to discover not so well performing pages or elements of your site and replace them.

The essence of the A/B testing is to understand in practice how the changes in some elements of the website pages influence the behavior of visitors. Let us assume that we decided to place the CTA in another part of the web page to see how it will affect the conversion. We conditionally divide the traffic into two equal parts. To the half of the visitors (group A) we show the old version of the web page where our CTA is located, for example, at the top right. To the other half (group B) – a variant where the CTA is right in the middle of the page. If we see that among the group B more people made a key action – then it’s obviously the best option.


In the same way any web page elements can be tested. Another thing is how reasonable is testing of various elements in the context of your website and your goals. For example, if you want to increase the number of people who clicked the button “buy” – think about the best arrangement of this particular element of CTA. If you need to increase the time a visitor spends on a page (especially relevant for Landings) – work on the layout and design of the page, namely on the colors, images, fonts, arrangement of elements. Want more subscribers – then consider what changes can be made in the form of conversion: test its appearance, location, number of fields to fill.

What else is tested: the headlines, the amount of text on the page, product descriptions, pictures, prices, availability and content of promotional offers, etc.

In order to quickly get the most specific and unambiguous results, do not just heap up many options (as is done in the A/B/n testing). Otherwise, you risk getting lost in the array of the received information, which requires complicated analysis. This is exactly the case when the simplicity is the key to success. Show a blue button to the group A and a red button to the group B. Did the red one play better? Excellent. Consider this option and test the next element. Now the group A will be shown the red button and the product description of a pair of sentences, and the group B will be shown the same red button, but a description on a whole paragraph. Further actions on the same principle. However, if you have a lot of traffic and enough time, you can apply multivariate testing. With it, you can test at once several elements in different combinations and find out how small changes affect the behavior of visitors.

How long is the testing carried out? At least until you reach the statistical significance – it is the required number of results that will give you the confidence in a result of 95% and above. As a rule, most testing services have a function of recognition of statistical significance, so you will be given the results for interpretation on time.

Popular tools for testing:

  • Content Experiments from Google – a free service for Google Analytics users that requires the skills of working with HTML‐code.
  • Visual Website Optimizer – a paid service that also requires knowledge of HTML.
  • Convert.com – a paid service for conducting A/B and multivariate testing.
  • Unbounce – a paid service that helps to create and optimize landing pages, as well as to conduct their A/B testing.
  • Optimizely – another popular paid service of A/B testing. Its main advantage is that the visual interface allows to test pages without digging in the HTML‐code.

Thus, the A/B testing is what will help you better understand the behavior of your website visitors and direct them to the necessary channel, and that actually makes it the essential thing for E‐commerce.

SoftwareGarage can help you to conduct split testing and many other E‐marketing activities. Just let us know about your project and we will think about the appropriate marketing tools for your business.

Also, subscribe to receive latest news and publications!

Newsletter

One Comment

  1. […] shipping or anything else. To figure it out you can use Google Analytics tool and carry out split testing. And the main problem – if a visitor leaves without buying. Although more often visitors […]

Comments are closed.