Goodbye Google Optimizer, Welcome Google Analytics Content Experiments
There has been a lot of new and exciting changes coming from the Google Analytics team lately. This Friday, June 1st, Google officially announced that effective August 1st, Google Optimizer is going away and is getting replaced by Google Analytics Content Experiments. Since a lot of our friends and customers will now be spending time trying to figure out what this change means to them, we thought we would offer some help by sharing our key findings and discussing some pros and cons.
First, please check out this YouTube intro video from Google Analytics team:
Here is our summary of the findings:
- A Google Account that has been associated with a Google Website Optimizer account is automatically enabled for Content Experiments.
- At this point we have not figured out a way to enable Experiments for Google Accounts that have an Administrator role on Google Analytics Accounts, but do not have a GWO account under the same email address. We will try to update this bullet as soon as we hear back from Google.
- Google will not offer a migration option for Google Website Optimizer users. You will have to re-create your GWO tests on the new Content Experiments.
- Only a control script for redirecting traffic from the original page to variations will have to be placed on the page that you are testing. No other tags will have to be added to the site to measure goals and conversions. All further tracking will be automatically completed by the Standard Google Analytics Tracking Code (GATC). Below is a screenshot of what the experiment code page looks like. Unfortunately, this code will have to be changed for each experiment.
- At this point there is no support for MultiVariate testing. This can be is a major limitation. However, knowing that there are some pretty smart people working at Google, I think this is just the first step towards integration of Testing and Analytics. We will be hearing a lot more from the Analytics Team over the next 6-12 months.
- We do not know much about it yet, but there is an improved statistical engine for analyzing experiments. It is supposed to make decisions faster. Definitely something we are going to look into quite a bit more.
- On the same note, Google states that with this engine, final test results will not be announced for 2 weeks, even though test results will be known very quickly.
- In the past, there have been cases when some “evil” marketers tried to abuse GWO by using it for cloaking purposes. With the new engine, the tests will automatically expire after 3 months.
- Another topic that we definitely need to learn more about is “Dynamic Traffic Allocation”. The traffic will be automatically switched from underperforming variations. This means that if your page is under performing (you are not paying attention or your Analytics Agency is not doing there job) the traffic will get automatically switched to a higher performing variation. Some obvious questions here is when this will happen, what would be the trigger for the switch, will it switch gradually, how will it allocate traffic given multiple winning combinations, etc… Don’t have the answer yet, but will try to provide an update as soon as possible.
- Another major benefit is the ability to leverage Advanced Segments and Page Metrics in order to segment test results and determine which pages work better.
- For now, Content Experiments have some hard limitations and it will be interesting to see over the next few months how the analytic community adopts these, and if Google will be forced to increase some of these limits:
- maximum of 5 variations per test
- maximum of 12 active tests per profile
On the positive side, there are some strong new features and a sign of a new architectural direction which is long overdue. On the negative side, a lot of these features are not going to work at advanced enterprise levels. It will be interesting to follow the news over the next two months (June and July) as the community starts to migrate from GWO and push back on some of the limitations. I know, I am an analytics geek, but I find this very exciting. How do you feel about Content Experiments?
Let me know by leaving a comment below: