In terms of e-commerce, marketing and sales, A/B testing can be carried out. This will result in successful and high-converting sales as you will gain a better understanding of why customers are not staying on your website and thus recognise which processes need to be improved. A/B testing is a powerful and effective tool for the company, making it easier to use and increasing the conversion rate.

Procedure for A/B testing

The target group of an A/B test, for example website visitors, is divided into two subgroups: Group A and Group B. This division must be randomised in order to obtain an average and even result from both groups. The areas/variables to be tested, for example the call-to-action button on a landing page, are then created in two ways: the current version and a new version. Two variables should only differ in one component, as this is the only way to clearly attribute a different reaction to a change.

You then use the original version of group Aand the modified version of group Bandcompare the reactions. The reaction in each case represents a desired effect, such as logging in, registering for a newsletter or ordering a product. The statistical test methods used in A/B tests depend on the characteristics of the data used, for which hypotheses are made, for example: "If the check-out process is shortened, the purchase completion rate will increase". A/B testing is a way to improve the user experience and significantly increase the conversion rate. This in turn has an impact on the company's turnover.

Requirements for A/B testing

In contrast to many variable tests, A/B testing only changes one variable and tests its effectiveness. A/B tests must therefore be easy to select in order to be meaningful and make the results valid. In particular, care must be taken when planning the test to ensure that the sample size is chosen randomly in order to reliably capture even the smallest detectable effect.

With A/B tests, it is also important to define suitable objectives and assumptions in advance so that success or failure can be measured later. They use two types of assumptions: Those where existing projects support the goals, and those that have not yet been implemented and have no numerical support but are reasonable assumptions.

Conversion optimisation using A/B Testing

A/B testing is a tool for implementing conversion optimisation strategies and should not be separated from other activities. As A/B testing tools test hypotheses statistically, you need to understand user behaviour in order to fully identify conversion issues. A/B testing should therefore be seen as part of conversion optimisation and not as a stand-alone tool. The conversion rate in online retail is less than two per cent. This is mainly due to the fact that conversion in general is a complex mechanism in which many factors play a role: the quality of the traffic generated, the user experience, the quality of the impressions, the popularity of the online shop or the activities of competitors.

The aim of A/B testing is to find and reduce the factors that could prevent a visitor from making a purchase on your website. With this test result, fundamental decisions can be made about possible courses of action. Various sources of information such as the time spent on the website or the percentage of shopping basket cancellations can provide valuable data for this. The data obtained describes how users interact with elements on or between pages. Even if not all of this data explains user behaviour, it can reveal problems with conversion, for example by identifying abandoned shopping baskets. They also help to prioritise the pages to be checked. Analyses with A/B testing therefore offer cost-effective insights into the website experience from the user's perspective and allow possible improvements to be derived later.

Tips for A/B testing

The SMART rule is used for a meaningful A/B test. With the help of a clearly defined problem of the target group, possible new solutions can be tested and later put into practice:

• Specific
• Measurable
• Achievable (measurable)
• Realistic
• In time

A/B tests initially require additional information in order to understand user behaviour. It must relate to a clearly defined problem with foreseeable causes. These defined problems should contain possible solutions to the current problem and indicate the expected results, which can then later be compared directly with the measured results of the users. For example, if the problem identified is a high bounce rate on an application form, the subsequent hypothesis could be: "If you shorten the form, the number of registered contacts will increase".

Factors in A/B testing

Which factors are measured with an A/B test depends on the problem in question. If you know that your website visitors do not understand the products you offer, you do not usually check the colour or placement of the "Add to basket" button. Instead, the product formulations are revised to suggest to customers what benefits they will derive from the products. This means that there are no standardised factors; instead, it must be tailored to the customer case. A possible purchase deficit may be due to the fact that the buy button is barely visible on the page. However, it may also be due to wording that does not convince the customer. This can be found out using A/B testing. The more precisely an A/B test is carried out, the more likely it is that the conversion rate will ultimately increase.