Introducing Website Testing in Google Analytics

by / Thursday, 14 June 2012 / Published in News

Testing, testing, and more testing is the mantra of many Internet entrepreneurs. If you have already set clear goals detailing what you want visitors to do while on your site, Google Analytics has introduced a method to test different landing pages to determine which converts best. Depending on the role your website plays in your business strategy, goals can include a visitor making a purchase, downloading a whitepaper, watching a video, signing up for a newsletter, or navigating to a specific URL. Content Experiments in Google Analytics allows you to present different variations of a landing page to different visitors and track metrics for each to determine which leads to the highest percentage of conversions.

This approach is different from A/B and multivariate methods of testing. Rather than two variations of a page available in an A/B test, or combinations of components on a single page, Content Experiments offers up to five unique pages each from a separate URL. With Content Experiments, Google Analytics allows you to compare how different versions of a landing page perform using a random sample of your visitors. You can also control what percentage of your visitors are included in the experiment and what type of goal you would like to test. Google made the announcement on June 1st and will be rolling it out gradually to users over the coming weeks.


Once the feature is available in your Analytics account, you will find it in the Standard Reporting tab, under the Content menu. When you open Content Experiments in the Content menu, you will see a list of your experiments after you have created at least one – otherwise you will be taken to a setup wizard overview page. To create a new experiment from this page, click Create Experiment – the setup wizard opens and asks you to complete four steps. First, you need to choose which pages will be part of your experiment by pasting the appropriate URL in either the Original Page or one of the Variation boxes. This will display a thumbnail image of the page so that you can be sure you have chosen correctly. In the second step, you will be asked to set the options for your experiment. This means choosing a goal you want to test for and the percentage of visitors that you would like to include in your experiment. The third step involves adding a piece of tracking code to the original page of your experiment set. Google Analytics gives you two options. Either you can copy and paste the code onto the appropriate page yourself, your you can have the code and instructions emailed to someone you are going to have do it for you.

Once this step has been completed, Google Analytics will test and confirm that the code is installed properly. The final step is to review your configuration and start your experiment. The Content Experiment setup wizard will automatically save the information you enter during the setup process, so you can return to complete your experiment set up at any time by opening the experiment list and clicking on the experiment you would like to continue building.

After you have setup at least one experiment, the Experiment List will act as the hub of information showing you which experiments are running and notifications about those experiments, start and end dates, and whether an experiment has produced a winning page. Clicking on an experiment in your list will bring you the the Experiment Report. This is where you can view detailed information about your experiment and metrics on how the individual pages are performing. You are also able to stop the experiment, change the visitor percentage, and disable particular pages from this report.

Content Experiments offers you a great way to test different versions of your web pages based on the conversion of a specific goal. By creating alternate landing pages, you can test headlines and headers, images and icons, text, calls to action, and page layout. The goals you can use in your experiments are somewhat limited in this first release. They include URL destinations and events including ecommerce conversions. Visit duration and pages per visit thresholds are not supported in this version, but more options are promised in the near future. Google Analytics makes several recommendations to act as guidelines to successful testing. They include testing only a few elements, using high-volume pages, making bold changes, and to do follow up testing.


While testing provides an excellent way to improve your visitors experience on your site and to increase conversions, it does present some potential SEO concerns which Google addresses. The first is cloaking. Cloaking involves presenting a search bot a different version of a web page than you show your visitor with the intention of deceiving the search engine to affect the page’s rank, and can cause your site to be sanctioned by Google. Creating different versions of landing pages that aren’t always shown to visitors has the potential to be abused as a cloaking device. Google spells out their guidelines for how to deal with this in the following statement in their help section:

To ensure that your experiments are not construed as cloaking, we suggest following the general guidelines:
Your variation pages should maintain the spirit of the content on your original pages. Those variations shouldn’t change the meaning of or your visitor’s general perception of that original content.
Keep your source code updated. An experiment ends when Content Experiments has collected sufficient data, and you should then update the source code on your test pages to reflect that winning combination soon afterwards.

We reserve the right to take action on any site that violates Google’s Webmaster Guidelines, or if we feel that visitors are being deceived or abused. For example, if a site’s original page is loaded with keywords that don’t relate to the combinations being shown to visitors, we may remove that site from our index.

Depending on the changes you experiment with on your variation pages, duplicate content issues may also arise. To be sure that duplicate content on your variation pages does not negatively impact your site’s ranking, Google suggests using the rel=”canonical” link attribute on these pages. This will signal to search engine bots that the variation pages are essentially the same and the original page should be indexed instead. If you will be leaving the variation pages on your site after the experiment has ended either to accommodate visitors who have bookmarked them or if you plan to use them for future experiments, it is recommended that you use a server-side redirect to point them to your original URL.

By incorporating a testing module directly into the Analytics interface, Google offers you an easy way to increase goal conversion and create a better experience for visitors to your website just a click away from viewing other metrics and reports so you can measure, test and optimize under one roof. They claim this is a step towards moving most of Google Website Optimizer functionality into Google Analytics. To learn more about Content Experiments, watch this video from the Google Analytics youtube channel: