How to 'Split Test' Your Website to Engage Online Customers Better
When it comes to designing your website, you have two choices: Select the site elements you believe look best, or you can use "split testing" to determine which design features are most engaging to your audience.
Essentially, split testing randomly serves up multiple versions of your website's pages to generate data about their effectiveness. The process allows you to quickly improve your website's overall conversion rates and isn't nearly as complicated as it may sound. Here's how to do it:
1. Isolate website page variables.
There are two types of split testing protocols: A/B split testing and multivariate testing. A/B split testing, which pits two page variations with a single differentiating element against each other, can be a better choice for beginning webmasters than the more complex multivariate procedure.
To get started with A/B split testing, you'll need to choose a single variable to test. A few possible options include:
- The wording, color, font, font size or position of your page's header text
- The specific colors used in your website's background or design elements
- The location of or wording used in your website's email newsletter opt-in box
- The position, color or design of your brand's social media buttons
- The color, design or location of any "Buy Now" buttons on your website
Related: How to Diagnose and Repair 'Conversion Rate' Problems on Your Website
The possibilities are endless so be sure to focus on elements that are likely to bring about the biggest changes in your website's performance. Tweaking the color of your links, for example, may not have as much impact as optimizing your on-site calls to action.
2. Build web page variations.
Once you've selected a specific variable to test, create the different pages that will be pitted against each other. Keep in mind that, for now, you'll want to limit your changes to a single variable to conclusively determine which elements result in better website performance.
If, as an example, you've chosen to test two different wordings for your site's headline text, you'll need to create a new page that will challenge your existing version. If your site runs in HTML, simply copy your original page, make the necessary changes to your headline text and then upload the new page to your website using a different file name (as in, "page-2.html").
Alternatively, if your website runs on WordPress or any other content management system, log in to your dashboard and create a new page using the same naming convention described above. Copy the text from your original page into your test page and then change the single test variable you're isolating with this A/B split test.
3. Install Google's Content Experiments program.
Once your test pages are ready to go, log into your Google Analytics account and find the instructions for installing Google's Content Experiments program onto your website.
This free program replaces the older Google Website Optimizer service, but still retains many of the same useful features. Although the Content Experiments program can let you test up to five pages against one another, most beginning webmasters will still find it easier to stick to a single test variable across two page variations. That can let you identify the specific elements that have the greatest impact on website performance as quickly as possible.
To set up a Content Experiments test, first select the pages to include in your split test. Then, specify to Google how much of your traffic should be exposed to the experiment, as well as the overall goal of your test. Most webmasters will want to set this value at 100 percent because doing so will help generate test data as quickly as possible. Finally, you'll be given a small snippet of code to install on your website, and your test will go live upon your approval.
4. Measure your results.
As your test proceeds, your results will be automatically updated within your Google Analytics account. However, there are a few things you'll want to keep in mind before making decisions based on this data.
First, it's important to wait to make any decisions about which page variation is the "winner" until Google Analytics is able to generate enough data to predict statistical significance. Judging the results of your A/B split test after only a few page visits simply won't give you enough information to draw solid conclusions about the best way to improve your website.
Once you do identify a winner, either move on to a new split test variable or create additional page variations to challenge your winning content. You want to maintain a mindset of continuous improvement for your website, so always be testing.