Run an Experiment

Configure & Modify

If you want to conduct experiments with your apps, you can use the Experiments API, or you can create those experiments in Tag Manager. You can also use the API for web-page experiments.

At any time, you can have up to 12 experiments per view that are running, in the process of being setup, or that are calculating your final data. Experiments that ended more than 48 hours ago do not count toward the limit of 12.

To create a new experiment:

  1. In Google Analytics, open the view in which you want to create the experiment.

  2. Click the Reporting tab.

  3. Expand the Behavior section, then click Experiments.

  4. If this is your first experiment, click START EXPERIMENTING.

    If you have already created other experiments, click Create experiment above the list of current experiments.

When you create an experiment, Google Analytics assumes that you have already taken time to prepare. That preparation includes identifying the objective of your test, choosing your original page, creating your variation pages, making sure the Google Analytics tracking code has been added to all pages, and determining the percentage of users you want to be included in your experiment.

1. Choose the objective

Enter a name for the experiment.

Under Objective for this experiment, select the metric you want to improve, for example, pageviews, transactions, or goal completions.

If you want to use goal completions and have not already created at least one goal, or if you want to create a new one now for the experiment, click Create a new objective.

If you create a new goal, click your browser's back button when you have finished, or click the experiment name in the list to continue with this step.

Under Percentage of traffic to experiment, select the percentage of users you want to include.

Remember that your original page is also part of your experiment. If you choose, for example, to include 50 percent of your traffic in an experiment with only one variation page, then 25 percent of your overall site traffic will see that variation page, and the remaining 75 percent will see your original page (half of the users included in the experiment, plus half of all your users).

Select whether you want to send email notifications about changes to the experiment. If you set this option to On, then use the menu to select addresses from your account users.

Click Advanced Options to set the following:

  • Distribute traffic evenly across all variations: Enable this option to assign an equal amount of traffic to each variation for the life of the experiment. When this option is disabled, Content Experiments follows the default behavior of adjusting traffic dynamically based on variation performance.
  • Set a minimum time the experiment will run: Set the minimum amount of time you want the experiment to run before Analytics declares a winner.
  • Set a confidence threshold: The minimum confidence threshold that must be achieved before Analytics can declare a winner. The higher the threshold, the more confident you can be in the result. A higher threshold can result in Analytics taking longer to declare a winner.

When you’ve finished, click Next Step.

Click Save for Later if you want to continue configuration of the experiment at a later time.

2. Identify the original and variation pages

In the Original Page field, enter the URL of the page you want to use as your control (for example, your existing home page, or the existing landing page for a product), then press Enter. If that page is live and publicly accessible, you see a snapshot of the page.

In the Name field just below the URL, enter a name for the original page so that it's easy to identify in your experiment report.

Consolidate experiment for other Behavior reports: Consolidate all traffic to your original and variation pages under the URL for the original page in your Behavior reports. This option lets you see all traffic to the URL of your original page. You can still see traffic to the individual pages (original and variations) in your Experiment reports.

What happens when URLs are rewritten

When you select this option, Content Experiments performs a head match on the variation URL, and replaces that portion with the URL of the original page. URL parameters are ignored. For example:

  • Original page: www.example.com/original_page

  • Variation page: www.example.com/variation1?id=50&cat=1

  • URL rewrite for variation1: www.example.com/original_page?id=50&cat=1

You can then specify variation pages, which should also be publicly accessible, by entering their URLs and giving them names. To add more variations, click + Add Variation near the bottom. To delete a variation, click the X next to that variation's name. You can include up to 10 variations of a page.

If your variation pages do not appear in the preview and you're sure they're available on your web server, investigate the known reasons for pages not appearing.

You may not see a preview for a page if it requires a login. In this case, you can follow the login request in the preview window, or load the page in another browser tab and follow the login request there. After you have logged in, refresh the preview window by clicking either the back or forward button at the top of the preview window, and then return to the page in question.

When you have identified all the pages you want to include, click Next Step.

Learn more about using URLs with dynamic parameters for your original or variation pages

3. Add and check experiment code

In this step, you have two options for adding the experiment code to your original page:

Manually insert the code

Content Experiments generates code to keep track of the users who see your original and variation pages. To add this code to your original page:

  1. Select Manually insert the code.
  2. Copy the code from the text box in the middle of the page.
  3. Open the HTML file for your original page.
  4. Paste the code immediately after the opening <head> tag, and before the Analytics tracking code (example below).

Original page without code

<html>

 <head>

  Contents of your header

...Google Analytics Tracking Code...

 </head>


 <body>

  Contents of your body

 </body>

</html>

Original page with code

<html>
<head>

 <!-- Content Experiments Control Script -->
  <script>

   ... The contents of your control script ...

  </script>
 <!-- End of Content Experiments Control Script -->


 ... Any other contents in your header section ...

...Google Analytics Tracking Code...

</head>
<body>

Contents of your body

</body>

</html>

Send the code to the webmaster

If you want your webmaster to add the experiment code to your pages, Content Experiments can send an email to your webmaster with all the necessary information.

To edit and send the email:

  1. Select Send the code to the webmaster.

  2. In the To field, enter your webmaster's email address.

  3. Click Send.

When your webmaster has added the code, return to the experiment by clicking the experiment name in the list.

If you’re comfortable with HTML and have access to your original page, you can add the experiment code yourself. Otherwise, you’ll need to contact your webmaster to help you add the code.

Remember that all the pages in your experiment need to have the Google Analytics tracking code installed.

The experiment code is synchronous, and is supported by all versions of the Analytics tracking code.

Check Your Code

Once you or your webmaster has added the experiment code to your original page, click Next Step to verify the code. At this point, Content Experiments checks to make sure the experiment code has been added correctly, and that each page contains the Analytics tracking code. For each page that is coded correctly, you see a green checkmark. If a page is not coded correctly, Content Experiments explains the mistakes and suggests remedies.

When you've successfully coded your pages, you can proceed to Step 4.

If validation for your variation pages fails and you're sure the experiment code has been properly added to your original page, it may be due to your web server not allowing unknown URL parameters.

4. Review and start your experiment

If Experiments has validated your experiment code and tracking code on your original and variation pages, and you are ready to start your experiment, then click Start Experiment.

If you want to start your experiment at some later time, click Save for Later.

If Experiments previously found problems with the code on your pages and you have resolved those issues, click Revalidate. Experiments makes another check of your pages. If Experiments displays error messages but you are certain you have your code configured properly, you can go ahead and start your experiment.

5. Modify Your Experiment

Up until the time you run your experiment, you can make changes to any of the options you chose.

To modify an experiment before you run it, click the experiment name in the list, and then click edit for any step in the wizard.

Once you have started an experiment, you can modify the experiment name, the names of the variation pages, the percentage of users who are served experiment pages, your notes, whether email notifications are sent, and you can disable specific variations. You can also validate the experiment code on your pages.

You can also copy a running experiment by clicking Copy experiment.

To modify a running experiment, click the experiment name in the list:

  • To validate that the experiment code is working properly, click Re-validate.

  • To disable a variation, click Disable Variation, then select the variation you want to disable. You cannot disable the original page.

  • To modify or copy an experiment, click Edit settings above the Explorer tab.

  • To stop an experiment, click Stop Experiment.