Blogger BlogNet14727

Multivariate testing can accelerate landing page conversion by 300% in one test by making it possible to evaluate multiple page elements at the same time.

If someone claims to be able to improve landing page conversion by this amount using other means, they are either very skilled at guesswork or lying.

The only full-proof method is by applying statistical techniques to the test process. By testing, I mean changing different elements that make up a web page and web site and testing to see if the conversion rate has increased or decreased.

This may sound straight forward enough but for anyone who has done even simple A/B split testing will know what a time consuming process it can be.

With simple split testing you could have to wait many weeks before being able to gather enough data from visitor actions and then use your own judgement or rule of thumb as to which version of the landing page proved to be better.

If you stop a simple split test when you think you have enough results and what you think is enough difference to show one page is better than the other, any mathematician will tell you the results may not be statistically significant.

This means you may not have gathered enough data and the differences in the resulting actions taken by visitors on each version of the landing page may not be enough to satisfy well established statistical sampling techniques.

Statistical sampling techniques have been around a long time and are applied routinely in the design, development, manufacture and quality testing of virtually all modern day products and processes.

Many of the current quality techniques and methodologies were developed for manufacturing by the Japanese who introduced such concepts as Kaizen meaning "change for the better" or "continual improvement."

In order to be able to test these quality improvements Genichi Taguchi, a Japanese engineer and statistician, developed a methodology and the mathematics to test the statistical validity of an outcome that improved the quality of manufacturing processes.

These same Taguchi Methods have been applied in different ways to the multivariate testing of web landing pages. The result is continual improvement to the landing page under test until the best quality conversion possible is achieved.

Unlike simple A/B split testing, multivariate testing allows you to test different page layouts, headlines, incentives, prices, images and colors, navigation and anything else you can think of and still discover the greatest improvement factors within a single test.

You are not limited to testing just 2 different headlines, 2 different sub-headlines, 2 different incentives or prices. It is possible to test many headlines, multiple prices, or tens of different incentives all during the same single test.

Multivariate testing solves the first major draw back with simple A/B split testing - the sheer time it takes to test so many elements and factors. Not only that, but typically 50% of all A/B split tests fail to give any enhancement, 30% show minor gains and only around 20% produce any significant improvement.

That means with a typical A/B split test lasting a month, if you conducted 1 test every month and 12 tests in a year, you can only expect 2 to 3 tests on average to produce any significant improvements.

With multivariate testing this is a thing of the past.

But in practice one would rarely test more than around ten changes at the same time since even with multivariate testing it takes longer, the more versions or elements that are being tested.

Multivariate testing also accelerates the improvement of landing pages by making decisions about which factors to stop testing.

Stop Wasting Valuable Traffic

Valuable traffic is therefore not wasted on testing page elements that have proven to contribute little or nothing and concentrate on factors that show statistically significant gains. This further reduces the time taken to reach a decision regarding the greatest improvement factors.

This combination of test factors brought about by the use of Taguchi test methods results in HUGE incremental improvements in landing page conversion rate and therefore in sales without needing large volumes of traffic.

Not only can multivariate testing analyze multiple factors in the same time, with the same volume of traffic required for one standard split test, but it can give you improvements in conversion rates that people using simple A/B split testing can only ever dream about.

Andy Theekson

Andy Theekson is the the Search Engine Optimizer a specialist who provides advice, tools and services to website owners wanting to gain more traffic from search engines, improve their visitor experience and increase sales. Andy provides a proportion of his advice for free and for a contribution to projects that help children in need. The continuing part of this article can be found at : Boosting Conversion Rates with Accelerated Testing