How to split test landing pages to increase conversion

In this post, I’ll show you the ins and outs of how to split test landing pages.

I’ll be talking about:

  • What split testing is
  • Why you should split test your landing pages
  • How the split testing process works
  • Forms of split testing
  • Website elements you can split test
  • The duration of a split test
  • How split testing can affect your search engine rankings

What is split testing?

Split testing (also known as a/b testing) is a form of testing where you compare two versions of a website to see which one performs better. Most of the time there’s just one little thing different between the website versions you’re testing so you can be more precise in finding out what creates a change in results.

Why you should split test your landing pages

Just because you’re getting a lot of website visits doesn’t mean they’re converting or that you have an optimum conversion rate. Split-testing makes it possible to make the most out of the traffic you receive on your website. It can be the difference between people not even clicking on your website and doubling the rate of your sales.

How the split testing process works

After hearing how much difference split testing can make when it comes to the conversion of your landing page it’s tempting to dive right in and randomly start testing every little thing on your website but its way more effective and takes way less time if you follow this process.

Study website data
Use analytics tools like Google analytics to find out if you have problem areas such as for example unusual high bounce rates or sessions being exceptionally short.

Analyse user behavior
Pinpoint what elements on the pages might cause this problem with the help of heatmaps, on-page surveys, visitor recordings and more.

Create a hypothesis
Based on the data assembled by looking at customer behavior you can start to form a hypothesis about where the problem might lie.

Test your hypothesis
Test this by making a change around that element. Example. Changing the color or text on a C.T.A button or changing an image you used for the landing page.

Analyze test data 
After making a different version of the landings page you send traffic to both versions to find out which one is performing better by looking at the amount of conversion each version has.

Apply results to your website
Make changes to your website based on the results of the split test.

Forms of split testing

There are multiple forms of testing that you can use to optimize the conversion of your landing page.
A/B testing
Where you make another version of the page with one difference and compare that to the original version.
Multivariate tests 
You again make another version of the page but it will have multiple differences that you can compare to the original version.
Multi-page funnel tests
Testing a funnel that goes over multiple pages and compares it to a different version of the multi-page funnel.

What can you split test

There are so many elements on a landing page that you can test that it sometimes is hard to even know where to start so I made a list of some elements you can check out.

– URL’s ( Short keyword rich URL’s rank better in Google)
– Headlines ( A benefit-driven headline that’s super precise is good for the CTR of your page.)
– Sub Headlines ( Sub headlines break up your text and make it more user-friendly.)
– Paragraphs text (The content itself is pivotal when it comes to conversions.)
– Testimonials ( Testimonials that are believable and relatable can do a lot for your credibility.)
– CTA text  ( What are you asking for and how are you asking it.)
– CTA buttons ( The location, size, and color of your CTA button can affect your conversions.)
– Color pallet ( The right color palette might be the difference between an amateur and professional look.)
– Images ( Confusing images can mess with your conversion.)

There are some more you can test out, but I just wanted to give an idea how many variables there are.

The duration of a successful split test

I can’t tell you how long you should run a split test but one thing I do know is that a successful split-test takes in almost every case longer than a day. To make sure you have a result that isn’t just a fluke you run the test over a longer period. Some people might even go up for 6 months just to make sure the results are accurate.

“You should not call tests before you’ve reached 95% or higher. 95% means that there’s only a 5% chance that the results are a complete fluke.” – Peep Laaja

How split tests affect your Search engine rankings

According to the Official Google Webmaster Central blog, Google is glad that people are testing their websites to increase performance, but it does have some implications. To keep these at a minimum they created a short guideline you can follow while testing your website.

No cloaking 
Cloaking is showing one version of a page to website visitors and another one to google bots. This going against the Google webmaster guidelines and as a result can cause your website to be demoted or removed from search engine results.

Use rel=“canonical”
Use rel=“canonical” is a link attribute you can put on the URL of all the alternate version of the website you’ve created for split testing. It lets Googlebot know that you prefer the original URL to be indexed. According to the GWMC blog, it’s better to use Use rel=“canonical” than meta-noindex, because in some situations it might pick one of the alternate urls as the one that should be indexed and the original URL might get dropped.

Use 302’s not 301 redirects

It’s also possible to run split tests by redirecting website visitors from the original URL to the alternate URL. It’s important in these situations to use the 302 redirect instead of the 301 redirect which is permanent. By doing this you’re communicating to search engines that the redirect is temporary so that they don’t pull your original page from their index. Javascript redirects are also allowed.

Only run the experiment as long as necessary

Every experiment has a different duration, but if Google deems your experiment unnecessary long they might see it as an attempt to deceive the search engine and they might take action accordingly.

Although split testing is a very useful tool to increase the performance of your website it’s not meant for everyone.
If you don’t have a lot of traffic it’s going to be hard to find patterns in the behavior of your visitors and you might risk coming to the wrong conclusions. If you’re new to this just focus on getting traffic to your website and take it one step at the time.

I’ve also created a post with the top 5 A/B testing tools you can use to easily modify certain elements of your website and test these different versions to see which one has the highest performance.

That’s all for now.

I hope this post was of value and if you have any question feel free to leave a comment.

Until the next post.

Leave a Comment