Articles in The 'GWO' Tag

Content Experiments in Google Analytics – Next Steps and what’s New?

by Theo Bennett

(This is the third post in a series about Google Analytics Content Experiments. For more information about Content experiments in Google Analytics (GA), you can view our first post two posts: Announcing Google Analytics Content Experiments and How to Create a Content Experiment in Google Analytics. )

So you’ve run your first content experiment in GA, now what? Well to answer that, we need to take a step back.

Hopefully your experiment began when you identified either an opportunity to improve your conversion flow or perhaps reduce the bounce rate on a landing page. Either way, your next step should have been to develop a hypothesis of why the page was not effective. Perhaps there was not a clear call to action, so you added a large button to encourage visitors to convert. Whatever you decided, your steps should have been

A. Identify a testing opportunity
B. Develop a hypothesis as to why the page is not effective and how it can be improved
C. Create a new version of the page and launch your experiment to test this hypothesis
D. Let GA determine which page is more effective

At this point there are three possible results:

1. You’re A page performed better than your B page
2. Each page performed about the same
3. Your new page, the B version performed better than the A page

Believe it or not, your next step is the same regardless of which result you saw. Let’s explore this. If you’re original page performed better or the same as the new one; that does not replace the original observation that this page/step/process could or should perform better. If your new page performed better than the old one then good for you; but that does not mean that it can’t still be improved. So as you may have surmised by now, the next step is to develop another hypothesis and run another experiment.

The good news is that there are new features available in Content Experiments to allow you easily pause, restart and re-launch saved experiments.

Happy Testing!

How to Create a Content Experiment in Google Analytics

by Theo Bennett

In my last post, I discussed the deprecation of Google Website Optimizer and Google’s Announcement of Content Experiments in Google Analytics. As a quick reminder, Website Optimizer is a testing tool that automates A/B and multivariate experiments. Content Experiments is an A/B testing tool that is integrated into Google Analytics. The benefits of such integration are many:

• Less code to implement

• Analyze and test from the same interface

• Easier to deploy (See the first bullet above)

Creating A Content Experiment in 7 Easy steps.

So you may be thinking, “Seven steps? In the GA interface there are only 4 steps!” And you are correct, there are four steps to creating an experiment, but there are three critical prerequisites.

1. As you have probably discerned, a Content Experiment is for testing another version of content. You will need to have already identified a critical page in your conversion funnel or an underperforming landing page to test.
2. After selecting your test case or “A” page, you need to have a hypothesis to test. For example, if you move the shipping cost calculations to this page/step in the checkout flow, more visitors will proceed to the next step and purchase your products.
3. Once you have your hypothesis of how to make your conversions jump, you’ll need someone (creative team, vendor, freelancer) to create the new design/ functionality etc. of the B page that incorporates your great new idea(s).
4. Once you have the prerequisites complete; drill into the GA Interface and go to Content>Experiments and enter the location of the page you want to improve.
5. Enter the location of the “B” version of the page that you’d like to test.
6. Set your options. Remember that you must have goals configured and the larger the audience, the faster you’re likely to have a winner declared.
7. Add extra, system generated coding to the “A” Page and have GA verify that all elements are in place.

So the final step is to launch your experiment. In our next post, we’ll explore some next steps and best practices once the experiment is complete. For now, you should know that once you’ve launched the experiment, GA will dynamically serve your “B” page instead of the original page at the rate that you’ve determined in step 6 above. Once the system has enough data to determine which page (A or B) is more effective; it will declare a winner and you’re on your way to more conversions!

Announcing Content Experiments in Google Analytics

by Theo Bennett

After your last log-in to Google Analytics (GA) you may have noticed that you have access to a new sub-section of the content reports. Google has launched Content Experiments by integrating an A/B testing tool into GA. Google also announced that Google Website Optimizer (GWO) will be going dark sometime in August 2012 as a result. Therefore, the rollout began with users that have run an experiment in Google Website Optimizer (GWO is an A/B and multivariate testing tool) and is continuing gradually to all users. If you are not sure whether you have this new feature yet, browse to Content > Experiments.

So is this a big deal? YES! If I’m not telling people they should always be segmenting then I’m saying you should always be testing. Content Experiments in GA allow you to run an experiment and improve your website performance right from where you do your analysis. Previously, you would identify a poor performing landing page in GA, for example, and then you would need to log-in to a testing tool like GWO to set up an experiment.

In addition to the unified analysis and testing that GA now provides; it’s also become easier to launch a test using a 4 step process that only requires you to have Google analytics coding on your site and a small extra control script on the A version in your test.

In my next post I’ll walk you through setting up an experiment, but for now you should know that you must have the control script on the A page as I mentioned above and a B version must also be live on your site for the experiment to launch. So you’ll also need a little help from your web team for the code and perhaps creative teams for the new landing page version. Start thinking of what you’d like to test and we will walk through how to tackle it.