Here’s something that should pique your interest in Google Analytics A/B testing.
I’m sure you’ve heard of the game The Sims-3. The makers wanted to drive more people to register for the game, and they felt that the value proposition was an area that needed some tweaking.
Let’s look at what A/B testing did for their business.
The original page looked like this (variation A):
Time to Dedicate Traffic for the Experiment
At this point you need to divide a portion of your website traffic for your experiment. This apportionment determines the ratio of people who will see the test pages as opposed to the original page.
If you’re expecting quick results, I suggest that you dedicate a high percentage of traffic towards the experiment. Of course, I wouldn’t suggest this if the experiment poses a few risks to your business.
Also, to remain in tune with the small changes that occur during Google Analytics A/B testing, be sure to turn the email notification on.
How to Distribute the Traffic
Click on the “advanced options” button. This will allow you to divide traffic by turning on the “distribute traffic” toggle.
Ensure that you direct equal amounts of traffic to each variation of the experimental pages. If you choose not to enable this option, the “Content Experiment” tool will go ahead and follow its default behavior. This involves distributing the traffic on an ongoing basis, depending on the variation performance.
For meaningful results, I recommend setting the experiment for a duration of at least three weeks.
The Confidence Threshold
This is why I consider Google Analytics A/B testing a trustworthy tool. It allows you to fix a confidence threshold for the A/B test. In other words, you can set a minimum confidence level that a variant page has to achieve in order to be announced as the winner. Not having any of the variants reach this threshold would mean all versions are a failure.
If you’ve set a high threshold, you’ll be more confident that the winning variation truly makes a difference. The point to be noted here is that a high threshold can require the experiment to be run for a longer period of time.
Configuring the Experiment and Code
It’s now time to configure the Google Analytics A/B testing Content Experiment. You’ll begin by adding your original web page and the experimental variants. Doing this is pretty effortless since all you need to do is add in the URLs as shown below. Then, take a look at the preview image for confirmation.
Once you’re done adding the original and test pages as well as previewing the images, click on “save.” You’ve successfully configured the experiment! Congratulations! But, you’ll now need to organize the experiment code.
Provided that the Google Analytics tracking codes have been installed properly on your variant pages and the original page, the codes would be visible in the box instantly.
Upon opening the head tag on the top of the original webpage, place the code immediately. Once that’s done click on “save changes” and proceed to the next step, which is the final one.
Review and Start the Experiment!
To make the whole exercise fool-proof, Google Analytics will validate the process once you’ve added the code. If any errors have been encountered, those will be highlighted. Sometimes, I’ve seen that Google Analytics fails to find the codes. In this case, you can skip the validation step.
However, ensure that this is only the last resort. Double check that your page is free of any errors before dismissing the validation step.
If all goes well, the Google Analytics A/B testing tool will show you the checkered flag which means the race begins. You’ll begin seeing the data within as little as two days.
Checking the Results
The experiment will run its course. At the end of it, Google Analytics will declare the winner.
The choice of the winner is going to be entirely based on the metrics that you set in place previously. Of course, the confidence threshold will also come into picture now.
Ideally, the process takes about three weeks, or a bit more than that. Eventually, you’ll have your “star” page right in front of you. This design will be the one that’s been most effective with your visitors.
When using the above outlined steps, you’ll hardly encounter any problem during the whole process, but I understand that you might need some extra guidance. I’m always here to answer your questions and lend a hand. Leave your comments below and I’ll help with your Google Analytics A/B testing. Happy experimentating to you!
About the Author
Shane Barker is a digital marketing consultant, named the #1 social media consultant in the nation by PROskore Power Rankings. He has expertise in business development, online marketing and is an SEO specialist who has consulted with Fortune 500 companies, government agencies, and a number of A-list celebrities.