Blog

A/B Testing With Google Analytics Content Experiments


A/B Testing, or split testing, is the best way to determine the best performing version of a web page. When you make a change to your website it should always be based on data. That is why testing different variations of a page with A/B testing is so important.

There are several different A/B testing tools available. I usually use Google Analytics Content Experiments because it is easy and free to use. You can test up to 10 variations of a landing page with Google Experiments. Just create the different page variations on your website, then set up the test in Google Analytics to randomly send traffic to each page variant.

Sometimes I prefer to use other A/B testing tools, like Optimizely, when changing small elements on a page. With Optimizely I can quickly change the color of a button without creating a new landing page on my website. But Optimizely is not free and can get pricey. I will discuss the pros and cons of Google Experiments vs Optimizely another time.

How to set up a Google Experiment

Step 1: Define your goal.

You should always define your goal before starting any A/B test. Ask yourself what problem are you trying to solve? What part of my page needs improvement? Some popular A/B testing goals include:

  • Increase click-through rate (CTR) on a call to action (CTA)
  • Increase form submissions
  • Decrease bounce rate
  • Improve session duration
  • Increase pageviews

Step 2: Create your test variations.

Now that you have defined your goal, it is time to set up your test variations. One variation of the page will always be the original version. You can create up to 9 more pages on your website that test different hypotheses and page elements. Here are some ideas of page elements to test to help achieve your goal:

  • Number of fields in a form
  • Page layout
  • Headline text
  • CTA colors and text

Step 3: Create your experiment in Google Analytics.

Log into your Google Analytics account and navigate to Behavior > Experiments.

Step 4: Choose an experiment objective

Fill out all the required fields in this first step of the Google Content Experiments setup.

  • Name Name your experiment whatever you want. Only you will see this.
  • Objective What did you decide was your goal in step 1? If it was Bounces, Pageviews, or Session Duration, you can find those goals under Site Usage. If you decided on a custom conversion goal, like a form submission, then choose that goal from your goal sets. If you have not set up your conversion goal yet, then do that now. For more information on how to set up custom goals click here.
  • Percentage of traffic If you want fast results from your a/b test, you should set a large percentage of your traffic to the experiment. However, if your test is risky and includes drastic changes, you should put a smaller percentage of your traffic to the experiment. Remember the original version will always be included in the percentage of traffic to the experiment.
  • Advanced options: distribute traffic evenly Google Experiments will automatically distribute most of your experiment traffic to the variation that is performing best for the set experiment objective. If you have more than one important objective and you want to give each variation a fair change, turn this option on to distribute traffic evenly.
  • Minimum time experiment will run Choose 3 days to 2 weeks for the length of your experiment. You can turn your experiment off at any time.
  • Confidence threshold Choose a threshold of 95% to 99.5%. The higher the threshold, the more confident you can be in the result.

Step 5: Configure your experiment.

Enter the URL of the original (control) page. This is the existing page on your website that you want to improve. Then enter the URLs of all the new variation pages. To add another variation, click the Add Variation link. Once you have entered your URLs for the original and variation pages, a preview will appear to the right of the URLs. If the page preview generates correctly and everything looks good, save your changes and move onto the next step.

Step 6: Set up your experiment code.

Once you have configured the experiment you can insert the experiment code onto the website. You can insert the code two ways:

  1. Manually insert the code paste the experiment code immediately after the opening head tag at the top of each experiment page. Do not add the experiment code to the top of every page on your website, just the experiment pages (including original).
  2. Send the code to your webmaster by email

Step 7: Start the experiment!

Once you have set up and validated your experiment code, you will see check marks on the review page. Click Start Experiment to launch your Google Content Experiment!

Step 8: Monitor your experiment

Now that your test has launched, you can monitor the results of your experiment by going back to the left-hand navigation in Google Analytics under Behavior > Experiments. As soon as data starts to come in, you will see a bunch of line graphs measuring the performance of each variation.

Step 9: Be patient!

It is important to remember that many tests will fail. Although it is easy to get discouraged when your tests dont work out as planned, dont give up! Use the findings from one A/B test to inform your next one.


Leave a comment

Your email address will not be published.