Last Updated on December 11, 2013
“Split tests” for pay-per-click campaign ads and their associated landing pages scientifically reveal the variables that generate the highest conversion rates. Split-testing tools within Google Adwords generate random ad configurations and collect data that predict each version’s conversion-generation success. You can then observe for yourself how tweaks to several variables will likely improve or hinder conversions.
True, you could arbitrarily compile your own preferred elements based on what your intuition tells you will attract clicks. Unfortunately, this eliminates the advantage of concrete data and leaves you with only your own judgment. Best of all, split testing takes very little time, effort or guess-work.
Split-testing both the ads and their landing pages ensures a perfect alignment at the most important stages in your campaign. Let’s examine a few steps.
- SELECTING THE VARIABLES
Split-testing your PPC ads and their linked landing pages works best in one of two ways
1. A/B TESTING
Tests focus on two versions of an ad or page, each altering a single variable – i.e., wording, color usage, wording within an e-mail newsletter opt-in box, social media buttons and positioning of other buttons
2. MULTIVARIATE TESTING
Suited to more advanced webmasters, this form of testing alters several variables in combination for deeper insights some variables even have their own sub-variables that can be tested. For example, other than the simple choice of words used in your landing page’s headline, you can also test different choices of text color, font, positioning of text and font size. Not every variable you test will impact projected conversions equally. Focus on the more significant elements improves your chances that you’ll stumble upon a major improvement.
- ESTABLISHING THE VARIATIONS
If you’re using Google Adwords’ split-testing tool to analyze your options, this is a fairly simple step that mostly involves following the utility’s prompts to create several versions.
If you’re split-testing landing pages, you’ll need to create your versions with changes only in the target variable (if A/B testing) or as many as you choose to test in combination (if multivariate testing.) Split-testing HTML involves copying your original code before changing the target variable and uploading the uniquely named new file to your site.
Using WordPress? Log in, create a new page as described above, copy your text and make the change you wish to test.
- MEET GOOGLE CONTENT EXPERIMENTS
This handy replacement for Google Analytics’ Website Optimizer can pit up to five versions of a single page against one another, for those feeling particularly confident.
Choose the page or pages that you’ll be testing. You’ll then detail your traffic-sample size and goals to content Experiments. After you install the generated code to your website that initiates the testing, the test goes live as soon as you give the word. In the end, you’ll receive deep insights into how changing various elements affect your site’s performance.
Remember, just because you can test up to five pages at once, doesn’t mean you have to examine so many. It remains simpler to simply pit one page against another and digest the test results one smaller fragment at a time.
- TAKE IN THE DATA
Once the tests have been completed, Google Analytics/Good Adwords will start updating the results. Just don’t be quick to judge what you see. Google Analytics/Adwords will need a little time before absorbing and processing data to conclusively point toward an optimal page with any statistical significance. Once you do have an undisputed champion, it’s your call: continue with a new split test that tinkers with a new variable or perhaps generate an entirely new page.