A/B testing describes a simple random experiment with two variables where the result of a new variable is tested against an existing variable. In the experiment, two random variables that are almost similar except for some minor differences are compared against each other. Supposing that the two random variables A and B, A was the older version while B was the newer version. The performance of B would be tested against the performance of A to see whether it is a better fit compared to A. A/B testing is also referred to as split testing, randomized controlled testing or an online controlled experiment. A/B testing is widely applied in computer programming, software development, marketing and web design among other areas.
An example of A/B testing is when new content of a website is tested against the older content version to see which one yields a higher conversion rate. The traffic to the website would be split into two with one segment experiencing the older version (A) and the other segment experiencing the newer version of the content (B) and the results of the experiment would be compared to show whether the newer version, B, is a better fit.
The process of A/B testing is carried out scientifically in the following manner.
First, identify the research question. This question forms the basis of the research and the answer to this question is the result of the experiment. For instance, the research question could be “why is the bounce back rate of my website higher that the industry standard?”.
The next step is to collect data. This can be done with the help of Google analytics tools which will provide the necessary information on your customers.
Using the information gathered, the next step is to draw a hypothesis. For instance, the hypothesis in this case could be “adding more links in the content could reduce the bounce back rate”.
The next step is to calculate the number of visitors per day that you will require to conduct your experiment after which you will test the hypothesis and then analyze data to draw conclusions.
Advanced A/B testing uses segmentation and targeting to see which version of a variable would yeild the desired results to a specific audience. The audience should be split based on specific attributes such as age, gender, occupation etc.
A/B testing is increasing in popularity with small and medium sized businesses who want to find out whether the marketing strategies in place are yielding results or whether they could be improved using other methods.