You may have heard of AB testing when you’ve come across information about web development. AB testing is when you design two versions of a web page (A & B) and divide traffic between the two. Then you determine which version has the most conversions. AB tests can help you determine what your conversion goal is and can help you determine what exactly is driving your audience to complete conversion.
However, although it seems simple, it can be very complicated to new web developers. Part of this reason is that there are many ways to test AB web pages. However, we’re going to show you a free tool that will help you test your AB web pages.
But before you can start conversion testing, you need to determine what conversion means for your business and website. Does it mean them signing up for a newsletter? Does it mean that they have contacted you through the website? No matter what your conversion goal is the end goal of conversion rate optimization (CRO) is to ultimately increase your business’ revenue stream through the website. But you must know what conversion means to you in order to identify the winning website in AB testing.
Once you know what your conversion is, you need to go to Google Analytics. Google has combined AB testing and split testing and calls it ‘content experiment’, which you’ll find in the Google Analytics page under Behavior and Experiments.
You’ll need to create an experiment and name it (typically a descriptive name is best so you remember what you’re testing for). You’ll also definite the metric you’ll be using to evaluate your results under the objective. Google Analytics has preset metrics and you will be picking from the list which includes: eCommerce, Adsense, Goals, and Site Usage.
If you’re confused about which one to choose, it’s easy to explain. Adsense is typically used to determine if the number ad clicks or impressions, if that type of conversion is your goal. If you’re looking to determine if the number of transactions on your website are growing, eCommerce is what you should choose. If you have your own predefined goals like session duration or destination page clicks, you should use the goal metric. And if you’re looking to see if the average page views or time on site is changing depending on the page, use the site usage metric. Fortunately, you can choose multiple metrics at one time, if you need to do that for your AB testing.
As soon as you’ve set your objective, you’ll need to divide your web traffic between your AB websites. This, obviously, controls how many people will be visiting your new test page compare to your original page.
During this part you can set how many visitors will be in the experiment. Higher numbers equal quicker results, but possibly also, less accurate ones. A smaller percentage may be better if your new page is quite different from your old one, in order to get a more accurate feel on the conversion rates.
There are also Advanced Options available on Google Analytics. There is an option that allows you control over how the traffic is divided, if you turn on the distribute traffic button. If you don’t use it, Google will adjust based on performance.
After you decide whether to turn on that option or not, you can set the length of the experiment. The longer it is left on the better, three or four weeks would be the minimum you’d want to leave it on.
You are also able to change the confidence threshold – the higher the threshold the more confident Google Analytics (and you) can be about the winning web page’s victory against the other design. However, longer thresholds also make your experiment last longer as more data will be needed in order to demonstrate a high confidence threshold. There are confidence calculators available online that can help you determine what threshold is best for your AB testing experiment.
You will need to configure your experiment by adding in the original web page and your new test pages. You will just enter the URL of your current page as well as the new one(s). Save your changes and move on to the next section, which refers to experiment codes.
Google Analytics codes should be properly installed on your original and testing pages after doing the above, so there should be an experiment code visible in the box on Google Analytics. You will need to “place this code after the opening head tag at the top of your original web page” and then, of course, save your changes.
The last, but certainly not least, step in Google Analytics will show Google Analytics validating the code you just added, and show you any errors (if applicable). If there are no errors and Google Analytics has found your code, then you will be ready to begin AB testing. The experiment will be launched and you should receive updates and data very soon.
Once your experiment has concluded, Google Analytics will “announce” the winner based on the metrics you defined as well as your confidence threshold. As you review your results, you’ll see which page worked best, and then you’ll be able to publish the new page, or keep the old one, with absolute confidence. There’s one fairly major con to Google Analytics, however, as Analytics doesn’t support testing more than one variable. This is a helpful experiment technique which allows you to test multiple variables like text size, location of buttons, color, etc. all at once. However, Google Analytics is still an awesome free tool that every web designer should use when performing AB testing.
At digiTech Web Design, we’re committed to offering our clients high-quality web design, SEO, and paid advertising services that are unparalleled. When it comes to designing web pages, we believe that Google Analytics is a great, easy to use tool to allows us to create and test the best webpages for our clients. For more information on how we can help you with your web design and internet marketing needs, please contact us today!
