Split Testing 101 [Increasing Conversions]

It isn’t a secret that when we want to make money on the Internet, we do things like make sales pages and opt-in forms to collect leads. But, just putting up these things is only the start. It takes a continual process of optimization to increase the percentage of your visitors that actually DO what you want them to do. This is where split testing comes to play.

If you have never heard of split testing, then now is the time to learn what it is and how it works. I’m not going to make you an expert in this post. This post is meant to serve as a primer and we’ll get into specifics later on.

A/B Split Testing

Essentially, split testing is simply testing one version of a page up against another version. You are comparing which of the versions performs better (i.e. more sales, more opt-ins, etc). Once you get a winner, then the top-performing page becomes your lead. You then repeat the process again. With continual split testing, you are constantly improving the conversion rate of your page.


This form of split testing is simple called A/B split testing. The name comes from the simple fact that you have two different pages, Page A and Page B, and you’re simply comparing the results of each. Obviously, when going A/B split testing, the results are most meaningful if the two pages aren’t completely different. After all, you want to know WHY one particular page converts better than the other. For this reason, you don’t want to change too many things at once. Perhaps just change your headline, a graphic, etc. By doing this, when results are showing that one page out-performs the other, you know why. If the two versions are COMPLETELY different, then there is a number of different things that could be affecting the results and you have no way to act upon the test results.

Multivariate Split Testing

The other type of split-testing (and more complicated) is called multivariate split testing. This is a split test where you are altering more than two elements in the same test (more than just pages A and B). Whereas A/B split tests just test one page against another, a multi-variable test can test any number of combinations. For example, you could test a number of different headlines, different images, different open paragraphs – whatever.

It works by dynamically inserting different test blocks into your landing page. For example, you would have your different headlines set up in your split testing system. In your landing page HTML, you would have some code which will dynamically pull in different headlines. Your testing system would then test all headlines and see which one performs better. If you are simultaneously testing other page elements (such as image placement), then the system will rotate these as well. Over time, the system will determine which combination of page elements converts best.

Multi-variable testing is more complex, but there is software to do it for you. Also, the more things you’re testing at once, the longer it will take to get meaningful test results. So, while you can load it up with all kinds of things, it is best to still limit your test elements.

Testing Software?

Picture 9 If you run a test for split testing software, you’ll get a lot of different options. However, for most people, I recommend you just use Google Website Optimizer. It is a service of Google and it is FREE. How cool is that?

Showing you how to use Website Optimizer is a subject for another post, however Google has done a pretty good job of dumbing it down. By going through the wizard they have set up, you can set up a split test on your site pretty easily.

Split Testing On a Blog?

Split tests are usually done on opt-in pages and sales pages. You don’t see too many people doing it on blogs. But, there isn’t any reason why you can’t. A few technical issues will need to be worked out, but it will work.

The thing to keep in mind, though, is that blogs are not always conducive to conversions. Blogs are great for publicity and building relationships with your audience. But, when it comes time for a sale, the blog can be a distraction. Ideally, pages designed to convert a reader into some goal of your’s should be simple and not cluttered up with a complex sidebar, banners, or anything like that. You want to force your readers to make a choice on whatever it is you got on offer.

If you keep that in mind, split testing gets a lot simpler. Trying to run split tests on an offer when your reader has a myriad of outbound links on your offer page is going to seem pretty fruitless.

So, more on this later! For now, this is a quick introduction to split testing. And I recommend you start playing around with Google Website Optimizer. It is a handy tool. Bookmark it. Any serious internet entrepreneur WILL use it eventually.


Sign up for free stuff. People still do that.

Join over 10,000 awesome folks who have become members. Your free membership means you can get help in our forums, access free stuff in the library, and you’ll get my exclusive newsletter (THE EDGE) in your inbox every Monday. For free.


  1. We split-test for our clients {convertify.io} but never thoght to split test with a blog. Very interesting and can’t wait to try it out!

  2. I am a huge fan of split testing (and everything that helps making decisions, honestly), but it is very interesting what you wrote about testing the blogs. I’ll take that into consideration.

  3. Nice read, very useful. Thanks. I tried using Google Web Optimizer, but wasn’t able to use it the way I wanted. I've now started looking at Logaholic Web Analytics [www.logaholic.com] to increase my website’s effectiveness. In combination with a lot of other in-built features they provide, it seems to do everything I need. Best thing is, it's easy to use and the reports are simple, clean and easy to understand.

  4. Nice read, very useful. Thanks. I tried using Google Web Optimizer, but wasn’t able to use it the way I wanted. I've now started looking at Logaholic Web Analytics [www.logaholic.com] to increase my website’s effectiveness. In combination with a lot of other in-built features they provide, it seems to do everything I need. Best thing is, it's easy to use and the reports are simple, clean and easy to understand.

  5. One important thing to remember, especially for people new to split testing is that you have to ensure that you have enough traffic coming to your site or blog to make your tests statistically relevant. Inaccurate assumptions based on low traffic volumes can get you in trouble and, in the long run, hurt your overall conversion.

    GWO – mentioned in the article above, does the statistical analysis for us internet Marketers that are NOT interested in dusting off our college stats books.

    Good post.

  6. I’m a real advocate of split testing. I’ve used simple A/B split test software to multivariate testing software… Google Optimizer is okay but I find that you get what you pay for and Google doesn’t do what some of the other software out there does… I personally like Paul Hancock’s power split tester for simple A/B split tests and either Split Test Accelerator or Evosplittester for multivariate testing with Split Test Accelerator being the Cadillac of the bunch… You have to test to know what works…

  7. found your site on del.icio.us today and really liked it.. i bookmarked it and will be back to check it out some more later ..

  8. I have heard about split testing, but never had a chance to try it. Is it possible for you to make a post/video on how to implement it in wordpress?

  9. Split testing is very important for increasing conversions online but so many people stay away from it thinking that it is too difficult. Great post!

  10. I think thats a really good explanation of split testing and really useful google tool and anything that can help you make money or more money is useful for you and also an insight into what people want to see from your site.

  11. So much to learn, so little time. I do love how much Google has available though. Has been a big help. Plus guys like David make it easier too.

  12. I’ve been doing that with some ads – testing the same ad with three slightly different titles, and the results are amazing.

    Out of 90 clicks, one ad got 60, the other 20 and the last one, 10.

  13. I love charts, graphs and stats. So this kind of testing interests me.

    It might be a problem determine whether the 2 separate pages have equal value to the visitors. If one page is inherently more interesting then another then the one page will get better results. Not because of the different layout or header size – but because it has better content.

    The only solution I can think of is to have a link that randomly goes to one of the two different layout pages. When a visitor clicks on the link they will be directed to one of the two pages. The pages have the same content, just the layout is different. This way you can make sure that you are testing just the layout and not the content.

    The only drawback is that if the visitor comes back and clicks on the link there is a chance that he will arrive at the other page.

    Great post – you got me excited. I think I will try it. All I have to do is create some code that randomly redirects the visitor to one of the two pages. Then I can make charts, graphs and analyze stats.

Related Articles You Might Like


I’ll send my exclusive weekly newsletter, THE EDGE. You’ll also be able to access exclusive content only found behind the login.
It’s all free. No obligations whatsoever.

Create Your FREE Account!

Join and participate in the community, access exclusive resources in the Document Vault, and get a free subscription to THE EDGE.

Your membership is free for life.