Multivariate split testing the easy way to bigger profits.

In it's simplest form split testing is also known as A/B testing i.e is 'headline A' better than 'headline B'?

This meant that before the computer became common office or home based business equipment, direct marketing involved the printing and sending out two different letters to a small selection of clients followed by manual updating of records to work out which headline was most effective. Then something else could be changed like the picture or colour of paper and another test was mailed out. As you can see, getting results would be slow at best.

Ignoring the fact that thanks to the internet we now have email,blogs and social networks which get our messages out instantly, combined with advances in programming technologies it now means we can get results of our tests almost immediately. What is even more appealing is that we can test multiple variations of a sales letter or process at the same time. This is known as multivariate testing. What's really impressive though, is we can now include testing multiple aspects of the adverts that lead to the sales page or funnel at the same time as part of the same process.

Multivariate testing is a process where various components of a website (or ad) can be tested. In simple terms this means numerous A/B or A/B/C/D/E/etc tests are performed on one web page at the same time. These tests are usually performed to determine which of your chosen variations works the best. In theory, multivariate testing can test the effectiveness of limitless combinations. In practical terms however, the only limits on the number of combinations or variables you decide to test are governed by the amount of time taken to get a statistically valid flow of visitors to your website. (It's also wise to make sure your server can handle the extra processing load if your site gets lots of visitors)

Taguchi testing was developed in the 1950's by Genichi Taguchi who is well-known for developing a methodology to improve quality and reduce costs, known in the United States as the "Taguchi Methods." He also developed the quality loss function.

These days Taguchi software and practices are used in conjunction with other technologies to design and perform highly efficient marketing experiments.

Taguchi testing means using a specific set of methods in a logical way to design and build test pages that your visitors see. Their response is recorded and analysed to increase the overall conversion rate (i.e., the desired action whether it be a sale or an opt-in).

Increasing conversion ratios will depend on many factors and the Taguchi software for advertising allows for accelerated testing. Essentially this means you don't need to run and validate all possible combinations of a landing page system which would be uneconomical. In short you can run all of the tests you want at the same time.

Using Taguchi software and methods systematically by choosing certain combinations of variables, you separate the individual winning contributions of each factor tested to the overall conversion process.

In short Taguchi testing and Multivariate testing can be thought of as one and the same thing. If you are using the right split testing software this is a very simple and intuitive process that can give you improvements of 200% and even more. But lets be realistic, even a 50% improvement would be worth having.

The simple answer is you can (and should!) test everything because until you do you won't know what on page factors give you the biggest improvements.

If you've read any books on writing sales copy you will know that the headline is considered by many to be the most important factor because if you don't engage and affect your visitor straight away they may well go elsewhere.

Then there is the sub-heading, the main picture and the first paragraph. With A/B testing you can test one against the other. With multivariate taguchi testing you could test as many (of each) as you like. (Imagine testing say 5 headlines with 5 subheads against 3 pictures).

But these are the obvious things that we should test. Here are a few others.

• Page background colour.


• Page width


• Font type


• The call to action


• Your bonuses


• The price - always a good one


• The colour of your headline


• With or without testimonials


• The number of testimonials - if with was better


• Main picture on the left or right


• With P.S. or without


• ?


• ?


• ?

The main point is - What works for me may not work for you.
More importantly never forget that a tested 'template' you develop that works on one of your sites may kill your conversions on another - even if it is in the same niche.

You can never test enough and if you haven't tested you really need to start now.