A / B test

from Wikipedia, the free encyclopedia

The A / B test (also split test) is a test method for evaluating two variants of a system in which the original version is tested against a slightly modified version. This method is mainly used in software and web design with the aim of increasing a certain user action or reactions. Over the years it has become one of the most important testing methods in online marketing . The A / B test is also used to compare prices, designs and advertising measures.

method

Ad optimization process using an A / B split test

In the A / B test, the target group (e.g. visitors to a website or recipients of a newsletter ) is divided into two sub-groups: Group A and Group B. This division must be random ( randomization ).

According to the division of the target group, the test object, such as a landing page or an advertisement , is also produced in two parts: the original variant and a modified variant. Both variants should only differ in one component, because only then can differences in the reactions be clearly attributed to the change. Then the original is used for group A and the modified version for group B and the reactions are compared.

The reaction here means the desired result, such as registration, registration for a newsletter or ordering a product. In addition to improving the user experience , A / B tests are therefore also a means of increasing the conversion rate .

Prerequisite / delimitation

In the A / B test, in contrast to the multivariate test, only one variable is changed and tested for its effectiveness. For an A / B test to be effective and the results to be valid , there must be a sufficient group size. In A / B testing, it is also important that corresponding goals and hypotheses are defined in advance so that the success or failure of a measure can be assessed afterwards. You work with two types of hypotheses: hypotheses that state that an existing element promotes the goal and hypotheses that have not yet been implemented and not substantiated with numbers, but are (apparently) logical. An example of a hypothesis would be: "A yellow buy button promotes the conversion rate". To get a clear result, one should always test hypotheses individually.

literature

  • Ash, Tim: Landing Page Optimization: The Definitive Guide to Testing and Tuning for Conversions. 1st ed., New York, NY: Wiley Verlag 2008, pp. 214ff
  • Kaushik, Avinash: Web Analytics: An Hour a Day. 1st edition, New York, NY: Wiley Verlag 2007, pp. 238ff
  • Schöberl, Markus: Tests in direct marketing: Concepts and methods for practice - evaluation and analysis - quality management and success orientation. 1st edition, Frankfurt / M .: Redline Wirtschaft Verlag 2004
  • Siroker, D., & Koomen, P. (2013). A / B testing: The most powerful way to turn clicks into customers. John Wiley & Sons.
  • Yoon Hyup Hwang (2019): Hands-On Data Science for Marketing: Improve your marketing strategies with machine learning using Python and R. Chapter 12.

Web links

Individual evidence

  1. ↑ Ron Kohavi, Roger Longbotham: Online Controlled Experiments and A / B testing. April 25, 2015, accessed February 8, 2016 .
  2. ↑ A / B test definition. Retrieved August 9, 2018 .
  3. ↑ a b Kreutzer, RT (2016). Online marketing. Wiesbaden: Springer Gabler. P. 141.