A fractional-factorial multivariate test is a design of experiment (DOE) where not all possible combinations of variables (each combination is called a recipe) are tested. Rather a sub-set of recipes is tested and a statistical approach such as the Taguchi method is used which makes simplifying assumptions. This DOE allows for a smaller sample and so speeds up the time needed to complete the experiment.
A fractional-factorial design relies on very small sample sizes (see Law of small numbers), and this reduces the flexibility and potential accuracy of the test design. Further, such tests assume there is no variable interaction between individual independent variables and this can lead to wrong conclusions being made. See also Full factorial.
Comparison of Full Factorial and Fractional Factorial Test design
Conversion marketing – Glossary of Conversion Marketing.
Over 300 tools reviewed – Digital Marketing Toolbox.
A/B testing software – Which A/B testing tool should you choose?