One of the great things about having a business website is that it makes testing variations of a page to improve your conversion rates relatively easy, however very few business owners actually do it! If you haven’t heard of split testing before, it is experimenting with two (or more) variations of a website page, and seeing which one compels your visitor to take the action you require, the best.
You don’t need to have a WordPress powered website to do this, although it does make it easier, but you will need a Google Analytics account.
I’m not going to go into how to create a Google Analytics account here, that’s for another post! Once you have setup your Google Analytics account you can start the process of setting up what Google calls an “Experiment”.
Firstly you will need to define what you goal for the page is. If, for instance you want to test a page to increase your newsletter sign up rate, the page will have a signup form on it.
You could test variations of the text compelling your visitors to sign up, or you could test the positioning of the signup form.
Let’s make this easy and say that the sign up form is on the top right of the screen in both versions and that we will initially just test the call to action.
In one version we’ll use the text “Sign up to our newsletter” and in the other, “Join our newsletter to get discount coupons and more”
Next, you will need to create (or have created) two or more variations of your test page.
Setting up your Experiment
Once you have your page variations created and uploaded to your server, you can go ahead and setup the experiment in Google Analytics. It is not immediately obvious how to do this in the analytics dashboard, so here’s how to do it.
When you are logged into the analytics dashboard, and selected the website that you want to run the test on (if you have more than one website on the account), look down the left hand menu and click on Behavior.
The menu will expand out and you will see a link called “Experiment”, click that link and you will see the following page.
Here, enter the website address of the page you want to run the test on. Then click on the blue Start Experimenting button.
Here you can give your test a Name. This is just for your information only so can be anything you like.
The Objective for this experiment can be set by either selecting one of the metrics from the drop down menu or you can create your own by clicking on the Create a new objective link.
This will take you to a new screen where you can set the goals, in this case you would create a new objective, that being increase newsletter sign ups. So, click on the Create a new objective link. The screen will change to this.
Click on the Create a goal button.
Here, you would select the last goal in the list, Sign up. Click Next step.
Here, to help analytics you will also need to create a thank you type page, thanking the visitor for signing up to your newsletter. When a visitor lands on this page, analytics will know which of the test pages brought the visitor there, and will log it as being the referring page. So on this screen, select Destination, then click on Next step.
The other two options can be left as they are.
Click on the create goal button.
When you have done this, make sure that goal recording is set to ON, on the next screen, you can then go back to the Experiment page, by selecting Behaviours -> Experiment from the left menu (you may have to hit the Home button at the top left of the screen to see the menu again).
Next, on the new Experiment setup page, you can select the Objective of this experiment drop down menu, and select the goal that you have just setup.
Next, you need to select the Percentage of traffic to experiment. There are a variety of percentage options available, but I would suggest you set this to 50% if you are only testing two pages. This way, the test will serve one page to 50% of your visitors and the other 50% to the other test page. Under the initial settings, you will see a link to Advance Options. Click on this to expand out the options, as below.
If you slide the Distribute traffic evenly across all variations, that is exactly what will happen, however, if you leave this off, it will be a bit more random.
The Minimum time for the experiment to run, can be set to 1 Day, 1 Week or 2 Weeks. Depending on the amount of daily visits you get to your website, would depend on which you would select. Obviously the more traffic your website gets, the accurate the results will be.
The Confidence threshold sets a minimum confidence level of the accuracy of the winning page. It is better to have this set relatively high to be sure that one particular page converts better than the other.
On this screen you will enter the URL of the variations of the pages that you are experimenting with. You can include other variations by clicking on the +Add Variation link. Upto 9 variations are allowed.
Click on the Next step button.
Here you can copy and paste the code into the head section of your experimental pages. This can be done either via the Appearance->Editor options (if you are familiar with changing the theme code, or if you are using a premium theme, there is normally a scetion that allows you to add the code to the header section without having to delve into the code.
Testing in WordPress
To experiment page variations in WordPress, you would do the exact same setup in creating two or more page variations. It is best to do this with pages rather than posts as posts may show up on a blog page, but pages do not have to show in menu’s.
There is also a plugin called Google Content Experiments that allows you to easily add the code to your page variations, rather than having to edit raw HTML pages.
I hope this helps you improve your website conversion.