How to Split Test – The What, Where and How Guide to Making More Money

How to Split Test – The What, Where and How Guide to Making More Money

(Today’s guest post comes from writer Mitch O’Connor, who correctly pointed out that, although I’m always telling you guys how important split testing is, I’ve never actually explained how to do it.  And since I’m still catching up from my vacation last week, I’m going to let him share a few thoughts on the subject…

Hope you enjoy, and please ask questions in the comments if there’s anything you want further clarification on!)

split testingAlthough SEO and web analytics attract a lot of attention among online marketers, they are not the only ways to increase your website’s conversion rates. By allowing you to observe the behavior of your visitors based on specified variables, split testing is one of the easiest and most direct ways to optimize your site. While the technique has been the subject of increasingly more buzz, some webmasters and bloggers are still unaware of what split testing is and how it can help them.

What is Split Testing?

Although there are two kinds of split testing, the most common is known as A/B testing. This type of testing involves creating two versions of a page, one of which has been changed in some way. Using a special application, you can test each page and determine which one is more effective at drawing in traffic, creating conversions, and earning profit.

Split testing can be done with nearly any part of a website, but it is important to only test a single element at a time to establish a control and make it easier for you to see how each change affects your website’s performance as a whole. Even small site adjustments can result in significant changes in profit and traffic. For example, a test conducted by Marketing Experiments showed that a page with slightly changed content and anchor text reading “click here” resulted in a sales increase of 15.57 percent. Doing such testing for every aspect of your website can increase this benefit even further. Numerous other tests and their results can be found at ABtests.com

Most of the time, changes made after split testing result in fairly low changes in conversion rate. A one percent difference might seem useless for a site that makes $100 a month (a gain of $1). However, if you can find 10 or 20 elements to change at one percent each, the modifications add up and can boost your site’s revenue dramatically (a gain of $20 per month, or a 20% increase in conversions). If you use split testing to increase your traffic as well, these numbers can quickly skyrocket.

Where to Split Test

You can split test nearly any aspect of your website, whether it be content, link placement, image size or image type. For example, let’s say you’re working on your affiliate relationships and your site is aimed at promoting Disney costumes from HalloweenExpress. You could host images of available costumes as links to the HalloweenExpress store, and then use split testing to determine which pictures result in the highest conversion rate. Likewise, you might test different headlines, product descriptions, pricing models and even color schemes.

You might think that minor page elements don’t affect the amount of visitors and profit you earn, but nearly every part of your site can benefit from split testing. For instance, the owner of web optimization and analytics blog empty mind discovered that replacing his old contact page button with an image of his own face boosted his conversion rate by 48 percent. This shows the importance of split testing as much as you can. You can never know how a minor change might affect your website’s performance until you actually test it.

How to Start Split Testing

Split testing might seem overwhelming if you’re not familiar with it, but the process is made easy with the help of specialized tools. One of the most popular ways to perform split testing is with Google Website Optimizer. This tool allows you to compare different versions of the same page against each other and provides you with concrete data to help you choose which version you should implement on your site. Various alternatives to Google Website Optimizer are also available, such as:

Visual WebsiteOptimizer
Webtrends Optimize
SiteSpect

To begin split testing, you need to create variations of the element you wish to test. For instance, you could come up with several different sign-up buttons to compare against each other. Once you have saved several different versions of your element within your testing tool, you can activate the test.

Once it has begun, the tool will select a different variation of your element each time someone visits your website. Following the test, the tool will provide you with details regarding the total traffic and conversion rate associated with each element. You could also make two separate versions of the same page and load them into the testing tool. Every time your site gets a visitor, the tool redirects them randomly to one of your two pages. The results the test returns should look something like this:

Image: LinkTrackr

About Multivariate Testing

If you’ve mastered A/B testing and want something even more powerful, you might consider learning about multivariate testing. This is a more advanced form of split testing in which multiple changes are tested at one time in order to achieve results and come to important conclusions more quickly. Instead of comparing two pages against each other, multivariate testing uses a single webpage, rotating between groups of elements in order to find the best possible arrangement.

Multivariate testing can be a quicker way to test your website, but it typically requires a large volume of traffic in order to provide accurate results. Although the results from a multivariate split test can be more difficult to interpret compared to its simpler counterpart, they can provide you with a wealth of useful information. Here’s an example of a multivariate test results page:

split testing 2

Image: Mandarich

To get a better grasp on this type of testing, Google has a tutorial video explaining how to set up multivariate testing experiments with Google Website Optimizer.

With all of these options in mind, do you currently use split testing (either A/B or multivariate) on your website?  If not, share the one thing that’s holding you back in the comments below!

About the Author: Mitch O’Conner is an online marketer and writer. When he’s not busy testing sites, generating traffic or writing content, he enjoys spending time with his wife and kids, watching TV, playing games or going camping.

Top Image: trekkyandy

2 Responses to How to Split Test – The What, Where and How Guide to Making More Money

  1. Sorry I’m late to the party again! I do use split testing on my website, but currently only for my opt-in forms. I’d like to ramp it up, but I’m not always sure what the best approaches to take are. Some of my questions are: Which things should I change and how should I change them? I know the purpose is to optimize, but I sometimes feel like there are too many factors to consider and adjust, making the whole task daunting.

    • Sarah says:

      Hey Jeffrey – That’s awesome that you’re already starting with split testing! My favorite elements to split test are:

      *Headline text
      *Different images and image placements
      *Style of checkout button (if selling my own product)
      *Text of call to action and checkout button
      *Anchor text to affiliate link
      *Different affiliate products
      *Different types of promotions (ie – in text links, videos, etc)
      *Placement of opt-in form and button text

      In general, I’ve found that these are the elements that lead to the biggest improvements via split testing (compared to something like background color, font face, etc), although obviously, not all of them will be relevant depending on the type of site you’re building.

      Let me know if you have other questions!

Leave a Reply