Google Replaces Website Optimizer with Analytics Content Experiments

Google Website Optimizer ending

Google just announced the launch of Content Experiments as part of Google Analytics. With this, they will be retiring Google Website Optimizer on August 1, 2012. For more details please see the Google Help articles on Content Experiments and the Analytics Blog annoucement. This is both good and bad for those of us working on conversion optimization.

The Good News about Content Experiments

Because this is part of Analytics, you'll have access to a richer dataset. It will be easy to create custom reports to isolate conversions. Consider the value of conversions - you can now quickly see how much revenue each variation has earned not just the number of conversions. Look at conversion rates by demographic data - has a variation shown an improvement in europe, but a decline in the US? Is there a relationship between browser versions or screen resolution and conversion rate?

Google apparently has improved the statistical engine. It's not clear yet what that means in practical terms. Will the improvements yield more accurate or faster results? The interface is a lot cleaner and easier to follow. Setting up experiments is easier. Content experiments natively allow advanced conversions like time on page, or analytics events.

The Bad News About Content Experiments

Multivariate experiments are gone, with no indication from Google whether they will be added in the future. This leaves us in a bit of a quandry. Multivariate experiments gave us the ability to perform site-wide experiments (e.g., testing "add to cart" buttons throughout the site). Experiments spanning multiple pages are gone as well. Google is limiting experiments to 6 variations as well, which forces more focused experiments (not necessarily a bad thing).

Google has not added Content Experiments to the Analytics API. This means there is no programmatic way to create, set up, launch, stop or even gather report data. If possible, we will still offer experiment import but only by attempting to find the information on a Google web page rather than a structured data feed.

We will keep you posted as we learn more about the changes and implications.