Free Webcast: Four Ways To Improve Your Content Marketing Maturity – Learn More ›
Popular Today in Business: All Popular Articles

Optimizing Website Conversion: An In-Depth Look

Online Marketing

Should your website be attractive? Functional? User-friendly? You should have answered “yes” to all of the above. However, a website can be all those things, yet still have a cringe-inducing conversion rate.

No matter how good it looks or how well it works, if your website doesn’t convert visitors into customers, it just isn’t doing its job. And if it doesn’t do its job, you’re missing out on opportunities and profits.

If your website already looks good and works well, how do you know what changes will improve your conversion rates? To figure out what’s keeping your visitors from becoming customers, you need a basic understanding of how split testing works and which tools you need to optimize your site.

Remove the Guesswork with Split Testing

Split testing is the best way to find out which change has the greatest effect on your website conversions. Split testing is the idea of splitting traffic to your site — with one half going to one version of the page, while the other half visits a different version of the same page. Whichever page posts a higher conversion rate is the winner.

Related Resources from B2C
» Free Webcast: Build Better Products by Identifying and Validating Your Riskiest Assumptions

Whether you test one idea at a time or make multiple, sweeping changes depends on your goals and experience. Although testing many ideas at once can help you make great strides in a short period of time, it can be difficult to tell which change had the biggest effect and how those changes are impacting one another. Testing one thing at a time makes it easier to understand how each change is working, but it can take much longer to achieve meaningful gains.

You have several options for split testing, and you should have a basic understanding of the three main types: A/B testing, multivariate testing, and multidimensional testing.

  • A/B testing: A/B testing is simple because there are only two choices: A and B. For example, you could test two different headlines for your homepage and see which one converts better.This can be done in two ways. You can create two different versions of the page hosted on separate URLs, and then enter the websites’ addresses into the testing software. The other option is to use paid software with a visual editor that allows you to make changes directly in the testing software. Although the second option is good for small changes, large, complicated changes are usually easier to deal with using the first method.

If you don’t have an in-house Web design team and are making changes on your own (or if you are just getting started in split testing), a visual editor is incredibly useful. The drag-and-drop format means you don’t need to know anything about code because the software updates the page for you.

  • Multivariate testing: In the conversion world, the word multivariate means “more than two.” Naturally, multivariate tests are just like A/B tests, except they have more than two versions or variables. An A/B/C/D test, for example, is simply four different pages with four different things happening. Each is being tested against the others at the same time. Multivariate testing, however, is not limited to four variables. You can run any number of pages against one another, such as A/B/C/D/E, A/B/C/D/E/F, etc. For this method to produce definitive results, you must have a decent amount of traffic coming to your site.
  • Multidimensional testing: Here’s where it gets really interesting. A/B testing or multivariate testing only tells you which page or variable performs better than the other. Multidimensional testing allows you to test all or some of the different variations in combination with one another. For example, let’s say you have a website with three different elements you want to test:

Idea 1: Changing the headline text.

Idea 2: Moving a button to the left.

Idea 3: Changing the background color.

The original headline, button, and background colors are labeled A, C, and E. The new headline, button, and background color are labeled B, D, and F. The ideas are then paired into combined sets:

A C E            B C E

A C F            B C F

A D E            B D E

A D F            B D F

Multidimensional testing allows you to see how all of these combinations work together. You can tell if two ideas work and one doesn’t, and then you can determine which one isn’t working. Unlike A/B testing and multivariate testing, you will know why the test fails or succeeds based on the changed and unchanged variables.

Testing and Implementing Changes with Optimizing Software

Unfortunately, the more complicated tests can get pretty confusing. Fortunately, there is software available that can do it for you. Here are some helpful tools to guide you:

  • Google Analytics Content Experiments: Google’s Website Optimizer has been one of the most common split testing tools for years. It has since been integrated into Google Analytics as Google Analytics Content Experiments. The power of split testing, combined with Google Analytics data, can be a fantastic tool for drawing conclusions and improving conversion rates. As you gain experience and your needs grow, you will probably want to switch to software with more features, but one of the biggest advantages of this software is that it’s free.
  • Visual Website Optimizer: Fast and user-friendly, Visual Website Optimizer is a robust piece of software geared toward visually-inclined users. Users don’t need to know any HTML or coding, nor do they need to do their own split testing.This software allows you to create variations, and then it splits the traffic, tracks the “winners” and “losers,” and declares an accurate winner for you. For example, the software might tell you that, based on statistics, one page has a 95 percent confidence level, which declares it the clear winner. More recent versions even launch the winner automatically so you don’t lose any money running a webpage with a lower conversion rate.

In addition to this quantitative data, the software also provides qualitative data, such as video charts that show which path the user followed to arrive at a certain point. Visual Website Optimizer also includes other advanced features, such as geotargeting and heat maps. The heat map feature allows you to see where people are clicking and interacting with your page and its individual variations. 

  • Optimizely: This is my favorite tool at the moment. The base functionality of Optimizely is very similar to Visual Website Optimizer. Although Visual Website Optimizer can be somewhat unwieldy, Optimizely was built to be much simpler. It has fewer features — heat mapping isn’t built in, for example — but the data is presented in an interface that’s more user-friendly. One feature allows you to set multiple goals (such as where people click or specific paths they take), and then watch as test results arrive in real time.

Interpreting the Results

After performing a split test, the most common action is to launch the winning variation. Every day that you don’t launch your optimized page, you’re losing conversions (money). Optimizely and Visual Website Optimizer both have a feature that automatically launches the webpage when the confidence level is high enough.

An important thing to remember is that you want to make changes in the areas with the biggest potential. Regardless of which type of split testing you use or which software you choose, try to take a holistic approach to conversion rate optimization. Think less about how the individual page converts to a lead, click, or sale, and look at the bigger picture. You can literally spend hundreds of hours optimizing one particular page when your time might be better spent focusing on a different page, optimizing your traffic, or following up with leads.

By using split testing to track the results of simple changes, you can learn what works best for your website and improve its conversions over time. You’ll be able to streamline, optimize, and focus your site, which can help ensure your company’s future success.

Please also view part one of this article titled, “Micro and Macro: Optimizing Website Conversions”

Comments on this Article: 2

Add a Comment
  1. Multidimensional testing is not a common word in the industry so lets not make this more complicated. The is the following A/B testing aka A/B/N test that is 2 or more variations (so 100 variations can be still an A/B test.

    Multivariate test is a combination of different sections. So what you call a Multidimensional experiment is and Multivariate test and what you call a multivariate test is called and A/B test.

    Check Visual Website Optimizer, Optimizely and our tool Convert.com. Lets keep is simple :-)

    Dennis van der Heijden
    CEO of http://www.convert.com

  2. Denis, Excellent point. Thank you for your feedback. Looking at it now, it appears my editor removed the multidimentional explanation from the article completely for the audience’s sake, but left in the title where it should have been a continuation about multivariate. But now, I feel as though we should do a followup article on just multidimentional tests since it is a more complex subset of a multivariate that we didn’t get to talk about here. I wish VWO or optimizely had a feature to make multidimentional testing easy. I am not sure if your product does or does not since its been a while since I had the opportunity to use it. If you are open to it, I’d love to find out more about where your product has evolved. Maybe we can even collaborate on some new feature possibilities.

Add a Comment:


Thank you for adding to the conversation!

Our comments are moderated. Your comment may not appear immediately.