Multivariate Testing

The multivariate test, also called A/B/n test, is the extended form of the frequently used A/B test. With the help of multivariate or A/B/n testing, more extensive shop adaptations can be evaluated for your benefit.

 

With multivariate or A/B/n tests, the variable “n” stands for the unlimited number of possible variants that can be tested against each other at the same time. For example, the color and placement of a button can be tested against each other at the same time. The users are randomly assigned to one of the different variants, as in a regular A/B test. However, this type of test requires a comparatively larger number of users in order to achieve a meaningful result.

 

The disadvantage of the multivariate or A/B/n test is that the analysis is based on an average customer. However, this cannot be defined as generally valid. Also, other deviations of the user cannot be considered. For example, regular and large customers can falsify the results of the test due to their purchasing behavior that falls far outside the norm. For this purpose, dynamic A/B/n tests are suitable. Here, the variants are repeatedly adapted on the basis of the user behavior.

 

Significance of an A/B/n test

 

A significance test indicates how likely it is that a result is not accidental and calculates the statistical error probability. Such significance calculators can be found both online and in various A/B testing tools.

Previous
Next
More Glossary Items
Demand Side Platform (DSP)
A Demand Side Platform (DSP) is an essential technology for automated trading of online advertising space....
Read More
Customer Data Platform (CDP)
A Customer Data Platform (CDP) combines data from different data sources into one system. The use of...
Read More
SEA
See also: Search Engine Advertising
Read More