As a webmaster, you know that it isn’t enough for your website to be “pretty” – it needs to be effective as well. Whether you’ve set up your site to sell products, collect leads or simply promote a certain viewpoint, your primary focus as a website owner must be to make sure that your site performs its designated function as efficiently as possible.
To do that, you need split testing – as in, the process of randomly serving up different page variations to viewers in order to conclusively determine which version helps achieve your target metrics. Here’s what you need to know about implementing this critical process on your own site:
Step #1 – Select a target variable
To get started with split testing, you’ll first need to select a variable that will be the focus of your test. This process is best explained with an example…
Suppose you have a page on your site that sells a digital download product. You’ve seen an unfortunately high bounce rate from visitors on this page, so you decide to use split testing to determine what changes can be made to keep visitors on the page longer and increase the resulting number of sales. For this reason, you decide to set up a split test that pits two different headline versions against each other, as this is the first element visitors to the page will encounter.
To conduct this test, you’d create two separate versions of your sales page – each one featuring a different headline. Once set up, your split test would direct visitors to one version or the other randomly, enabling you to determine which headline variant results in the most sales.
This specific example utilizes a type of split testing known as A/B testing, which takes its name from the fact that you’re comparing one page version against another in which a single variable has been changed. If you’d changed multiple variables on your test sales page (for example, your headline, your call-to-action and your “Buy Now” button color), you wouldn’t be able to conclusively determine which single action led to increased conversions.
A more advanced type of split testing – multivariate testing – is available if you’d like to be able to compare multiple variables at once while still generating meaningful data. However, implementing this type of test correctly is more complicated, which makes the process better suited to advanced webmasters, rather than beginning split testers.
Assuming you’re going ahead with the A/B testing protocol, you’ll want to identify your target variable before moving on to set up your test in Step #2. Besides your page’s headline, a few other elements you can test include:
- The placement and selection of images on your site
- The specific wording used in your calls-to-action
- The price of your products
- The location and wording used on your email list opt-in box
- The shape, color and location of your “Add to Cart” buttons
Really, the possibilities are endless when it comes to choosing a test variable. For your first test, try to select a variable that stands to make the biggest possible difference in your site’s performance (for example, test your headline text, rather than the font you use in your 3rd paragraph). Then, create the test version of your split test page and move on to the next step in this process.
Step #2 – Use split testing programs to launch your test
By far, the most popular program used to run split tests online today is Google Analytics’ “Content Experiments” (formerly known as the Google “Website Optimizer”). The program is free and incredibly simple to use, making it a great place for beginning marketers to start.
To run an A/B split test using Content Experiments, you’ll need three pieces of information:
- The URL of your original page (for example, sales-page.html)
- The URL of your test variation (sales-page-2.html)
- The Google Analytics goal you’ll use to determine which page variation results in a conversion.
Once you have all three of these items identified and setup, head to the “Content -> Experiments” section of your Google Analytics dashboard and begin by entering the URL of your original page into the prompt. Click “Start Experimenting” and you’ll be prompted to enter a few different pieces of information, including your test URLs, goal selection and a few other testing parameters.
Finally, after this information has been entered and verified, you’ll be provided with a new version of your Google Analytics tracking code that must be installed on all the different pages that will be involved in your test. If you’re technically savvy, you can add this tracking code on your own and then use Google’s built-in tools to verify that it’s been added correctly. If not, Google Analytics provides a helpful set of instructions that can be sent to your site’s web developer.
Step #3 – Identify a winner and launch a new test
After completing the verification process and launching your test, you’ll see data start to flow in to your Google Analytics dashboard as visitors arrive on your site and interact with your pages. Eventually, Google Analytics will report a “winner” for your experiment, based on the confidence threshold you set during the setup process (the default is 95% confidence).
Keep in mind that it’s very important to wait until this measure of statistical significance has been reached, as picking a winner after only a handful of goal completions doesn’t provide you with any meaningful data. If your site is young and not yet receiving many visitors, it can take awhile to reach your requested confidence threshold, but be patient! Making decisions off of incomplete data is no better than failing to split test in the first place.
Once a winner has been chosen in your test, make any necessary changes to your live site based on your results and then launch another test immediately. The most effective websites rely on a program of continual split testing to make much-needed improvements. Don’t short-change yourself by running one test and then quitting.
That, in a nutshell, is the A/B split testing process, as well as how it can be used to make your website more effective than ever. If you need further information, check out the Google Analytics “Overview of Content Experiments” help section for more detailed instructions on using this program to improve your website results.
