Google Website Optimizer is gone.¬† It has been replaced with Google Content Experiments.¬† Countless articles and blog posts have been written on how this change will affect testing for everyone.¬† This blog post will briefly outline why Google Content Experiments will make your life better, but might make your life harder, and will show you some alternative testing tools that this change should draw your attention to.
Regular Expressions Filter Out IP’s
RegEx is an awesome new feature because you can easily include or exclude specific IP addresses from a test.¬† Now you won’t dilute your test results every time you view your tests pages because you can exclude yourself and the people you work with from being counted in the experiment.
Content Experiments are Integrated into Google Analytics
Since Content Experiments are now integrated with Google Analytics, you won’t have to spend so much time wondering, “Why don’t the numbers in my GWO account match what I’m seeing in GA?”
No more wasting valuable time trying to figure out or explain small inconsistencies that drive you crazy.
Experiments Have to Run for 2 Weeks Minimum
Google Content Experiments will not confirm statistical significance until a test has been running for two weeks.¬†¬† There are different reactions to this change.¬† Many sites feel they have seen enough traffic before the end of two weeks to be sure which page performs optimally.¬†¬† While we agree that it is a pain to run a test longer than you feel you need to, it is more important to be sure that your test results will be lasting.¬† There is too big a chance that something outside the landing page itself could effect the outcomes in any given week.¬† Imagine an ideal test-it would run for years to confirm significance across days of week, seasons and fashions throughout the years.¬† We think two weeks minimum is an appropriate compromise.
We know that you might see things differently and want full control over the abilities of your testing tool.¬† It is probably time for you to start looking at some alternative testing platforms.
You Can Look at How Your Test Pages are Performing by Segment
The addition of Advanced Segmenting is the very best part of Google’s new Content Experiments.¬† Now you can look at test outcomes based on traffic type, device, whether or not the visitors to the page have seen the page before- the insights are endless.¬† Most importantly, you can see that one page might actually convert better overall, but not fair well for a specific segment.
You Don’t Need A New URL for Every Test Variation
This upgrade¬† allows you to test different page versions without having to create multiple URLs.¬† You can create a version of your test page that has a different URL parameter using a symbol like # 0r ?.¬†¬† Using relative URLs will allow you to test different versions for a product template page on your Ecommerce website.
They Took Away Multivariate Testing!
Google tried to make the transition away from multivariate testing easy by allowing you to use relative URL’s in Content Experiments.¬† The thing is, it just isn’t the same. Multivariate testing made it so much easier for highly trafficked sites to see how multiple changes to a page influence and affect one another.¬† The inability to test how elements on the page interact to affect conversion rate is a serious loss.
You Have to Keep Losing Variations through to the end of the test
Although Content Experiments will try and shift traffic away from low performing variation(s), it doesn‚Äôt give you the ability to pause them.¬† If you started with 5 variations and only two are performing, you used to be able to pause the under performing pages in GWO to conclude the test more quickly. In Content Experiments, you will either have to start the test over, or allow the test to run longer than it needs to because of the presence of the diluting variations.
In this industry where time is money, Content Experiments falls short with this change.
You can only have 5 variations and 12 active tests running at any given time
We cannot imagine a case when you would ever have 12 tests running on your site at once.¬† There are too many factors in each test that will affect one another and you won’t be able to find out anything from your results.
We do not see the purpose of restricting the number of variations running per test at any given time though.¬† If you have enough traffic, and more than 5 optimized variations- you should be able to run them all at once.¬† It doesn’t seem likely that you will ever have more than 5 variations that you think might be the best, but if you do- you should be able to test them!
Visual Website Optimizer- This is one of our favorite testing tools.¬† It is very easy to use, the price is very reasonable and you can track multiple goals at once.
Optimizely- No coding required. A one-time copy-and-paste is all you need to run tests. Only works with A/B testing
Omniture- An all inclusive analytics platform.¬† One of the industry favorite analytics and testing platforms
SiteSpect- A/B testing, multivariate testing, targeting and personalization, mobile content optimization as well as web performance optimization all without having to implement page tags or change your site in any way.
Unbounce- Test, create, publish and optimize landing pages without IT.
Convert Experiments (previously known as Reedge)- Seamless Integration with Google Analytics Goals, E-Commerce Revenue and Conversion Tracking
Webtrends Optimize- An all inclusive analytics platform.¬† Includes advance segmenting, mutivariate and A/B testing.
What is your favorite A/B or multivariate testing tool?¬† Please share your experiences and insights with other readers so that they might have some idea of what new testing tools they might like to try out.¬† Have you started using Google Content Experiments yet?¬† Do you love the integration with Google Analytics?¬† Have you come across any frustrations from the change?¬† If you need help setting up tests on your site, we can help from creating variations to test to making sure your tests are implemented properly.