Tum Dena Saath Mera - 3rd January 2012

What Is Split Testing? Let Madonna and Lady Gaga Explain

lady

This article will explain what is split testing is with the help of two of pop music's most popular divas. Although there seems to be no obvious link between your website's sales and these singers, there is an interesting parallel there.

Understanding this parallel will enable you to increase the ROI of any sales page without doing any extra marketing.

Madonna has reigned supreme over the world of pop for the last 30 years, not only with her catchy songs but her numerous image reinventions. However, it's now clear that someone is stealing her crown: Lady Gaga. While we still marvel at Madonna's fitness and showmanship, nothing can top a meat dress for sheer outlandishness. The younger singer connects with her fans (or 'monsters' as she affectionately calls them) in a way that Madonna never has.

While Madonna has gone through many image makeovers in her career, Lady Gaga seems to reinvent herself every month. Yet Madonna wins on longevity every time. She was making hit records long before Lady G was even born - songs that we still play today.

So what is split testing, and what have these singers got to do with it?

This form of testing is where you create 2 or more versions of a webpage. Version 1 will differ from version 2 - you then measure which version gives you the best sales, most sign ups (or whatever the criteria for success that you have for that particular webpage).

There are 2 types of split testing to be aware of:

A/B Split testing is where you just want to test one element of your webpage: for example in your headline, will the phrase "Double Your Profits" bring in more customers than "Improve Your Profits"?

It contrasts with multivariate testing, which is where you test multiple elements of the webpage: so you might alter not just the headline, but the leading image, the font, the layout...the list goes on. So version 2 of your webpage might look radically different to version 1. Not so with split testing - in the example above only the word 'Improve' would change.

We can look at A/B split testing as Madonna and multivariate testing as Lady Gaga.

Madonna's image changes are relatively well-paced and laboriously orchestrated (say to coincide with a new album: think dirty chic for Ray Of Light in 1998 or disco diva for Confessions On A Dancefloor). In this way they resemble split testing.

Lady Gaga's image changes seem far more chaotic (although they, too are probably very well planned) - she's wearing a telephone on her head one minute, and a self flaming bra the next. Just like a webpage in a multivariate test, one image change can look drastically different to the next.

The $ 64,000 question is: which is better? A/B testing or multivariate?

The answer is that it depends - the two have very different functions.

Use A/B split testing when you want to know why one headline (or layout, or image etc) outperforms another. You can then apply this knowledge to other sales pages - you might well find that you always get a higher ROI with "Double" rather than "Improve" in your headline.

Use multivariate testing when you just want to improve your results. Say that sales page version 2 (with its new headline, font, image and special bonus) increases sales by 25%, we cannot pinpoint exactly why this is. However if the company needs the extra sales as a priority, this is obviously more important than knowing why turnover has increased.

The Lady Gaga multivariate model is therefore a short-term tactic, and the Madonna split testing model a long- term growth strategy.

Neither Madonna or Lady Gaga is inherently a better performer than the other - it all depends on what you want. We can say the same about split or multivariate testing - it all depends on what you want to achieve with your sales page. The next step is, of course, to decide exactly what changes to test in your sales page.