Detonate Conversions Using Multivariate Split Testing

August 26, 2011 by  Filed under: Marketing 

Multivariate split testing can easily boost web page conversions by up to 300% or more. One single test can be used to appraise numerous web page components simultaneously.

There are those who claim to be able to enhance conversion without testing but If an individual claims to be capable of improving webpage conversion they’re either very good at guessing, plain lucky, or they are resorting to misinformation.

Split Testing Exposed

Really the only idiot-proof technique is the use of statistical methods in the split testing process. By testing you will be modifying various elements that comprise a website and testing to determine if the conversion rate has grown or dropped.

This might seem straight forward enough however anyone who has ever done even basic A/B split testing knows what a time intensive procedure it can be.

With basic split testing you might have to wait patiently for weeks before you are able to gather enough data from visitor actions to make a difference and then use your own judgment or rationale as to which version of the webpage turned out to be better.

In the event you stop a basic split test once you think you’ve got enough results and what you consider is sufficient difference to show one page is preferable to the other, any math wizard can tell you the results might not be statistically significant.

Which means you may not have collected sufficient data and the variations in the resulting actions taken by website visitors on each variation of the webpage might not be sufficient to satisfy well-established statistical testing methods.

Statistical sampling methods have been in existence for a long time and they are employed routinely in the style, improvement, production and high quality testing of pretty much all modern day products and procedures.

Most of the existing quality techniques and methods have been developed for manufacturing by the Japanese who introduced these kinds of concepts as Kaizen.

To be able to test these quality advancements a Japanese engineer and statistician, designed a technique as well as the math concepts to test the validity of an end result enhancing the standard of production processes.

These same methods have been used diversely in the multivariate testing of web pages and sites. The result is constant improvement to the webpage being tested until the highest quality conversion possible is reached.

In contrast to basic A/B split testing, multivariate-testing enables you to test various page elements simultaneously, yet still identify the greatest improvement factors in a single test.

With this method you’re not limited by testing just 2 different elements of your page. You can test anything that you like on your page using one individual test.

Multivariate testing resolves the 1st significant draw back with basic A/B split testing, the amount of time it takes to evaluate so many elements and components. Typically about 50% of all A/B split tests aren’t able to provide any improvement, 30% show minor increases and just around 20% produce any substantial improvement.

This means using typical A/B split testing lasting one month, in the event you conducted 1 test each month and 12 tests annually, you may only assume two or three tests generally produce any substantial advancements.

Multivariate testing relegates simple A/B split testing to a relic of a by gone era.

Though there is something that you should take into consideration. Testing more than 10 different elements at the same time will significantly increase the amount of time that it takes to complete a test.

Split Testing and Traffic

Visitors are not wasted on testing page elements that have proven to add little or nothing while focusing on factors that demonstrate statistically significant gains. This further decreases the time taken to achieve a decision concerning the best improvement factors.

This mix of test factors brought about by using multivariate methods leads to increased conversion rate and as a by product, sales, without the need for a lot of traffic.

Multivariate testing can evaluate many factors at the same time, with the same amount of traffic needed for one common split test, it can present you with enhancements in conversion rates that those using basic split testing could only dream about.

J. Allen Reeves is a conversion rate specialist who does conversion rate optimization for his clients, who have made millions of dollars in extra profit because of his strategies. To get a FREE calculator to discover how much money you’re currently losing due to poor conversions, go to

Article Source:

Speak Your Mind

Tell us what you're thinking...
and oh, if you want a pic to show with your comment, go get a gravatar!

You must be logged in to post a comment.

Prev Post:
Next Post: