Responsive website design, Norwich
 
 
 
 

Google Content Experiments

by Mike Amis

Google Content Experiments

Content experiments are a simple and effective way to optimise website landing pages for your chosen goals.

The objective of most web pages is to encourage visitor interaction in some way, for example, fill in a form, subscribe to a newsletter or make a purchase. Content experiments make it possible to test different variations (up to 10) of landing pages to see which one is the most effective at improving the metric you are optimising for.

This approach is different to the A/B split test model where small variations on the same page are tested such as the wording of a call to action, this A/B/N model allows for up to 10 completely different page variations to be tested using different URLs, as long as all of the variations are served from the same domain name.

With content experiments it is possible to:

- Test multiple page variations

- Choose what percentage of your visitors are shown page variations

- Choose which analytics goals you would like to optimise for

- Get email updates about your experiment whilst it is in progress

Setting up experiments

Before you can access content experiments you will need a Google Analytics account. We create an analytics account and add the tracking code as standard for all websites we build so you will have access to insights and opportunities from day one.

Once you are signed in to your Google analytics dashboard, from the lefthand menu select:

REPORTING -> BEHAVIOUR -> EXPERIMENTS -> START EXPERIMENTING

Google analytics experiment set up step 1


Here you need to name your experiment, make sure you choose a meaningful name that you will recognise in the future. From here you also need to choose a goal objective for the experiment. This could be a goal you have already defined or one that is built into google analytics such as bounce rate. You also get the opportunity to set up a new goal from here.

Next you select the percentage of your total traffic that will be involved in the experiment. The original version of the page is also included in this percentage. It is worth bearing in mind that whilst it can be tempting to include all of your traffic in the experiment, if your variation pages perform worse than your original you will actually see a downturn in goal completion whilst the experiment is running.

The next two options set how long to run the experiment and what degree of certainty in the result you want before declaring a winner. A higher degree of certainty will require more experiment sessions and therefore take longer to achieve but the upside is that you can have more confidence in any changes you make to your website based on the result.

Configure experiment pages

Configure google content experiment pages

Here you define the URLs of the pages used in the experiment and name them so you can easily differentiate between them in the reports. Checking the box 'Consolidate experiment for other content reports' will add all of the experiment traffic together when you view other reports whilst the experiment is running so you can easily still see total traffic volumes. Traffic is still separated in to the different variations when viewing experiment reports.

Add experiment code to your website

With all of the options and variants for the experiment set Google will generate a Javascript code snipped that needs adding to each of the experiment pages. There are two options here for adding the code, to do so yourself or to invite your web developer do this for you. Adding the code requires some knowledge of HTML and it's position on the web pages is important, so unless you are experienced it is best left to your developer.

Once the code is added you need to validate it from the Google Analytics dashboard in the next step before you can start the experiment. When the experiment details are correct and the code appears correctly on the page, you can start the experiment. After 24-48 hours of data collection you will be able to monitor the progress of your experiment in a report. If you need to end the experiment before a result has been found, you can do so by selecting the experiment name from the list and clicking STOP EXPERIMENT at the top of the page.

Google analytics experiment set up confirmation

Share this post
Leave a comment