HomeGlossaryMultivariate Testing

Multivariateadj : having or involving a number of independent mathematical or statistical variables. (http://www.merriam-webster.com/dictionary/multivariate)

 

For marketing professionals, multivariate testing refers to the practice of isolating different components of a marketing campaign to determine the influence each has on success. Multivariate experiments are extremely useful because they necessitate manipulations of content, headlines, colors, etc. to measure what element variations are most desired (or preferred) by a target audience. This method of testing offers key insight for future marketing campaign improvements.

Examples of Multivariate Testing

A common use of multivariate testing in a marketing setting is email. The ability to test several subject lines and/or layouts helps discover the best voice to use for intended readers, as well as the ideal structure to deliver email content.

Another common use case is testing variations of calls to action (CTAs) on a piece of content. Depending on the audience of that content, a reader may respond better or worse to a specific ad variant. You can see an example of this in the image below.

This image shows a method of multivariate testing by showing different ad versions on a piece of website content.(Click the image to view its full size.)

Other examples:

  • Search Engines: testing different link colors and fonts to catch more clicks. (CNET)
  • Paid Search: Testing variations of PPC ads to see which delivers the best click-through rate and conversion.
  • Twitter: manipulating the text of a Tweet and retweeting it over time to see which version draws greater engagement.
  • Direct Mail: creating a variety of different postcard designs to send to prospective clients and measuring which has the highest response rate and overall quality.
  • Airline Industry: prices are dynamically altered based on a custom algorithm and presented to pools of queriers, seeking to maximize revenue for the airline in each situation.